The Pirate Post ha ricondiviso questo.

Four Chained Flaws in Microsoft SCCM Let Any Domain User Seize Full Server Control
#CyberSecurity
securebulletin.com/four-chaine…
The Pirate Post ha ricondiviso questo.

The media in this post is not displayed to visitors. To view it, please go to the original post.

Il filesystem /proc su Linux: la guida completa per il troubleshooting da sistemista
#tech
spcnet.it/il-filesystem-proc-s…
@informatica


Il filesystem /proc su Linux: la guida completa per il troubleshooting da sistemista


Ogni sistemista Linux, prima o poi, si trova davanti a una domanda che top o free non riescono a risolvere del tutto: perché questo processo consuma tanta memoria? Perché una porta risulta occupata ma netstat non mostra nulla? Perché un servizio continua a scrivere su un file che, in teoria, è stato cancellato? La risposta, quasi sempre, si trova in un posto che molti amministratori conoscono solo superficialmente: /proc.

/proc non è una directory come le altre. È un filesystem virtuale, di tipo procfs, generato e mantenuto interamente dal kernel in memoria: non un singolo byte dei suoi contenuti risiede su disco. È, di fatto, una finestra live sullo stato interno del kernel, esposta attraverso strumenti che ogni sistemista già conosce: cat, grep, awk. Comandi come free, ps, top e iostat non fanno altro che leggere e formattare i dati che /proc mette a disposizione: imparare a leggerlo direttamente significa avere accesso alle stesse informazioni, ma senza filtri, e con la possibilità di combinarle in modi che nessun tool preconfezionato offre.

La struttura: due mondi in una directory


/proc contiene sostanzialmente due categorie di contenuti. Da un lato le directory numerate, una per ogni processo in esecuzione (/proc/1234/, dove 1234 è il PID), che espongono lo stato specifico di quel processo. Dall’altro i file di sistema globali, come /proc/meminfo o /proc/cpuinfo, che riflettono lo stato del kernel nel suo complesso, indipendentemente da quale processo li legga.

I file di sistema che contano davvero

/proc/cpuinfo e /proc/stat


/proc/cpuinfo elenca i core logici della CPU con modello, velocità di clock, dimensione della cache e i flag di supporto per estensioni come vmx (virtualizzazione hardware), aes e avx2. Un one-liner molto usato per contare le CPU logiche disponibili è:

grep -c '^processor' /proc/cpuinfo

/proc/stat, invece, espone i contatori di tempo CPU aggregati e per singolo core, suddivisi in colonne: user, nice, system, idle, iowait, irq, softirq. Su un server sotto carico I/O-intensivo, tenere d’occhio la colonna iowait nel tempo è spesso più diagnostico di qualunque dashboard grafica, perché permette di isolare il momento esatto in cui la CPU comincia ad attendere lo storage invece di lavorare.

/proc/meminfo: oltre a “free”


/proc/meminfo è la fonte dati grezza dietro al comando free, ma contiene molti più campi di quelli che free mostra normalmente. Due meritano attenzione particolare: MemAvailable, che indica la memoria realmente disponibile per nuovi processi tenendo conto della cache riclamabile (a differenza di MemFree, che è quasi sempre un numero fuorviante), e Dirty, che riporta quanta memoria contiene dati modificati ancora in attesa di essere scritti su disco — un valore che cresce pericolosamente su sistemi con storage lento e scritture intense.

Per un monitoraggio in tempo reale della pressione di memoria, questo comando è estremamente utile in fase di troubleshooting:

watch -n1 'grep -E "MemAvailable|Dirty|Writeback|SwapFree" /proc/meminfo'

/proc/diskstats, /proc/loadavg, /proc/net/


/proc/diskstats fornisce le statistiche I/O per dispositivo — letture, scritture, settori trasferiti, tempo speso in operazioni — ed è la fonte dietro iostat. /proc/loadavg espone le tre celebri medie di carico a 1, 5 e 15 minuti, mentre /proc/uptime riporta i secondi trascorsi dal boot e il tempo cumulativo di inattività della CPU.

La sottodirectory /proc/net/ merita un capitolo a sé: /proc/net/dev contiene i contatori di byte e pacchetti per ogni interfaccia di rete, /proc/net/tcp e tcp6 elencano tutte le connessioni TCP attive in formato esadecimale, /proc/net/sockstat riassume l’utilizzo dei socket a livello di sistema, e /proc/net/snmp espone contatori SNMP utili per individuare retransmit e fallimenti di connessione anomali.

Il mondo per-processo: /proc/PID/


Qui si trova la parte più preziosa per il debugging quotidiano. Alcuni file chiave:

  • /proc/PID/cmdline — il comando esatto con cui il processo è stato lanciato, con gli argomenti delimitati da byte null.
  • /proc/PID/status — un riepilogo leggibile con UID, GID, utilizzo di memoria e numero di thread. Il campo VmRSS indica la memoria residente effettivamente occupata, mentre VmSwap segnala se (e quanto) il processo è finito in swap.
  • /proc/PID/fd/ — una directory di symlink, uno per ogni file descriptor aperto dal processo. È la base dati su cui si appoggia lsof. Contare i descrittori aperti è semplice: ls /proc/12345/fd | wc -l.
  • /proc/PID/io — la contabilità I/O del processo, con read_bytes e write_bytes che riflettono i dati effettivamente passati al layer di storage.
  • /proc/PID/maps e /proc/PID/smaps — le regioni di memoria mappate dal processo (codice, stack, heap, librerie condivise). smaps va oltre e fornisce, per ogni regione, i dettagli su RSS, PSS e swap.
  • /proc/PID/environ — le variabili d’ambiente al momento del lancio, anch’esse delimitate da byte null.
  • /proc/PID/cwd e /proc/PID/exe — symlink rispettivamente alla directory di lavoro corrente e all’eseguibile su disco. Sono fondamentali per identificare binari che sono stati cancellati o sostituiti mentre il processo era ancora in esecuzione (un classico segnale da cercare dopo un aggiornamento di sicurezza mal gestito).
  • /proc/PID/oom_score e oom_score_adj — il punteggio che l’OOM killer del kernel usa per decidere quale processo terminare per primo in caso di esaurimento memoria. oom_score_adj, con un range da -1000 a 1000, permette di proteggere o “sacrificare” volontariamente un processo specifico.

Due scorciatoie tornano utili spesso: /proc/self/ punta sempre al processo che sta effettuando la lettura, mentre /proc/$$/ punta alla shell corrente all’interno di uno script.

Casi pratici di troubleshooting


Trovare quale processo ha in ascolto una determinata porta (qui la porta 443/0x01BB su IPv6), senza usare ss o lsof:

inode=$(awk '$2 ~ /:01BB$/ {print $10; exit}' /proc/net/tcp6)
ls -l /proc/*/fd/ 2>/dev/null | grep "socket:[$inode]"

Individuare processi che tengono aperti file già cancellati — una causa classica di dischi che risultano pieni pur senza file visibili, perché lo spazio non viene liberato finché il file descriptor resta aperto:
find /proc/*/fd -ls 2>/dev/null | grep '(deleted)'

Ispezionare i contatori di interrupt per CPU, utile per diagnosticare squilibri di IRQ affinity su macchine multi-core:
cat /proc/interrupts | head -20

/proc/sys e il tuning via sysctl


Sotto /proc/sys/ si trovano i file scrivibili che espongono i tunable del kernel — la stessa interfaccia usata da sysctl. Ad esempio, per modificare a caldo la propensione del kernel allo swap:

echo 10 | sudo tee /proc/sys/vm/swappiness

Tra i tunable più comuni: vm.swappiness (0-200, propensione allo swap), net.ipv4.ip_forward (abilitazione dell’IP forwarding), net.core.somaxconn (backlog massimo per i socket in ascolto) e fs.file-max (limite di file descriptor a livello di sistema). Ricordiamo che le modifiche fatte così sono volatili: per renderle permanenti serve intervenire su /etc/sysctl.conf o su un file in /etc/sysctl.d/.

/proc dentro i container


Un punto che genera spesso confusione: /proc è namespace-aware. Dentro un container Docker, la directory mostra il namespace PID del container, non quello dell’host — il PID 1 visto dall’interno è l’entrypoint del container, non l’init dell’host. Attenzione però: campi come quelli di /proc/meminfo o /proc/stat spesso riflettono ancora i totali dell’host e non i limiti imposti dai cgroup al container, un dettaglio che ha tratto in inganno più di un tool di monitoring scritto ingenuamente.

/proc contro /sys: quando usare cosa


/sys (sysfs) è il filesystem virtuale complementare a /proc, ma orientato al modello dei dispositivi del kernel: driver, bus, block device, gestione dell’energia. Un esempio tipico è /sys/devices/system/cpu/cpu0/cpufreq/, che espone i parametri di scaling della frequenza per singolo core. La regola pratica: /proc per processi e stato del kernel a livello di sistema, /sys per configurazione e stato dei dispositivi hardware.

Conclusione


Conoscere /proc non è un esercizio accademico: è uno degli strumenti diagnostici più potenti a disposizione di un sistemista Linux, proprio perché sta sotto ogni tool di monitoring che si finisce per usare quotidianamente. Quando top, free o netstat non bastano — o semplicemente non sono installati sulla macchina compromessa o minimale su cui ci si trova a lavorare — sapere leggere direttamente /proc/meminfo, /proc/PID/status o /proc/net/tcp fa la differenza tra un troubleshooting rapido e ore perse a indovinare. Vale la pena tenere a portata di mano, come riferimento minimo, i file /proc/meminfo, /proc/cpuinfo, /proc/stat, /proc/net/dev e la struttura di /proc/PID/: coprono la stragrande maggioranza dei casi reali di debugging su sistemi in produzione.

Fonte originale: A Guide to Linux /proc Filesystem, LinuxBlog.io


The Pirate Post ha ricondiviso questo.

The media in this post is not displayed to visitors. To view it, please go to the original post.

Crittografia post-quantistica: perché il threat modeling non può più aspettare (con esempi in .NET 10)
#tech
spcnet.it/crittografia-post-qu…
@informatica


Crittografia post-quantistica: perché il threat modeling non può più aspettare (con esempi in .NET 10)


C’è una minaccia alla crittografia moderna che non richiede un computer quantistico funzionante oggi per essere reale oggi: si chiama “harvest now, decrypt later” (raccogli ora, decifra dopo). Un attaccante con risorse sufficienti — uno stato-nazione, tipicamente — può intercettare e archiviare traffico cifrato con RSA o ECC adesso, per poi decifrarlo tra qualche anno, quando computer quantistici sufficientemente potenti saranno disponibili. Per dati con un ciclo di vita lungo — segreti industriali, cartelle cliniche, comunicazioni diplomatiche, chiavi di firma di lungo periodo — la finestra di esposizione è già aperta, anche se il computer quantistico “rompi-RSA” non esiste ancora.

È in questo contesto che Microsoft ha recentemente pubblicato una guida al threat modeling applicato specificamente alla migrazione verso la crittografia post-quantistica (PQC), invitando organizzazioni e team di sviluppo a non aspettare la disponibilità degli algoritmi per cominciare a muoversi.

Perché serve il threat modeling, e non solo un elenco di algoritmi


Il problema centrale che Microsoft evidenzia non è tecnico in senso stretto, è organizzativo: la maggior parte delle aziende non ha un inventario completo di dove e come viene usata la crittografia nei propri sistemi. Certificati TLS, librerie di firma incorporate in applicazioni legacy, hardware embedded con chiavi hard-coded, protocolli proprietari che usano RSA “perché si è sempre fatto così” — sono tutte dipendenze crittografiche nascoste che un aggiornamento generico non intercetta.

Il threat modeling applicato alla PQC serve esattamente a questo: mappare asset, flussi di dati, confini di fiducia (trust boundary) e controlli di sicurezza esistenti per far emergere queste dipendenze prima che diventino un problema urgente sotto pressione regolatoria o, peggio, sotto attacco. Concretamente, significa rispondere a domande come: quali sistemi cifrano dati che devono restare confidenziali per più di 5-10 anni? Quali certificati di firma del codice hanno una validità che si estende oltre il 2030? Quali protocolli interni non supportano l’agilità crittografica, cioè non permettono di sostituire un algoritmo senza riscrivere l’applicazione?

Gli algoritmi: da NIST a nomi che iniziano a essere familiari


Dopo anni di standardizzazione, il NIST ha pubblicato tre standard che è ormai il momento di conoscere, perché stanno entrando nei prodotti che usiamo quotidianamente:

  • ML-KEM (Module-Lattice Key Encapsulation Mechanism, FIPS 203) — sostituisce lo scambio di chiavi basato su RSA o curve ellittiche (ECDH). È l’algoritmo che entra in gioco nell’handshake TLS per stabilire un segreto condiviso.
  • ML-DSA (Module-Lattice Digital Signature Algorithm, FIPS 204) — l’algoritmo di firma digitale post-quantistico, pensato come sostituto di RSA e ECDSA per firmare certificati, codice e messaggi.
  • SLH-DSA (Stateless Hash-Based Digital Signature Algorithm, FIPS 205) — un algoritmo di firma alternativo, basato su funzioni hash anziché su reticoli, pensato come backup conservativo nel caso in cui in futuro emergano debolezze crittanalitiche nei problemi su reticolo.

Sul fronte delle raccomandazioni pratiche, Microsoft insiste su un punto spesso sottovalutato: la migrazione a TLS 1.3 è il prerequisito, non un dettaglio. TLS 1.2 non supporta i meccanismi ibridi necessari per introdurre ML-KEM accanto agli algoritmi classici, e restare su TLS 1.2 significa restare bloccati fuori dalla transizione PQC anche quando gli algoritmi saranno pronti. In parallelo, si consiglia di alzare l’asticella anche sulla crittografia simmetrica e sulle funzioni hash, adottando AES-256 e SHA-384 come standard, per mantenere margini di sicurezza adeguati anche in scenari quantistici (dove alcuni attacchi, come Grover, dimezzano di fatto la sicurezza effettiva degli algoritmi simmetrici).

La timeline che Microsoft si è data per la transizione dei propri prodotti e servizi critici è il 2029 — una data che vale la pena tenere a mente come riferimento di settore, anche per pianificare le proprie migrazioni interne con lo stesso ordine di grandezza.

Cosa c’è già oggi: le API disponibili


La parte più interessante per chi scrive codice è che la PQC non è più solo teoria da paper accademico: le API sono già disponibili e generalmente disponibili (GA) sulle piattaforme Microsoft.

Windows: CNG


Su Windows, il supporto arriva a livello di CNG (Cryptography API: Next Generation), con identificatori di algoritmo dedicati per ML-KEM e ML-DSA utilizzabili tramite le funzioni BCryptGenerateKeyPair e le relative API di key exchange e firma. Il supporto nativo è arrivato a partire da Windows 11 con gli aggiornamenti di novembre 2025: chi gestisce flotte Windows dovrebbe verificare di essere allineato con le patch più recenti prima di pianificare qualunque test.

.NET 10: MLKem e MLDsa in System.Security.Cryptography


Per chi sviluppa in C#, la notizia più concreta è l’arrivo delle classi MLKem e MLDsa direttamente in System.Security.Cryptography con .NET 10. Ecco un esempio minimo di key encapsulation:

using System.Security.Cryptography;

if (!MLKem.IsSupported)
{
    Console.WriteLine("ML-KEM non è supportato su questa piattaforma");
    return;
}

MLKemAlgorithm alg = MLKemAlgorithm.MLKem768;

using (MLKem privateKey = MLKem.GenerateKey(alg))
using (MLKem publicKey = MLKem.ImportEncapsulationKey(
    alg, privateKey.ExportEncapsulationKey()))
{
    publicKey.Encapsulate(out byte[] ciphertext, out byte[] sharedSecret1);
    byte[] sharedSecret2 = privateKey.Decapsulate(ciphertext);

    bool match = sharedSecret1.AsSpan().SequenceEqual(sharedSecret2);
    Console.WriteLine($"I segreti condivisi coincidono: {match}");
}

E un esempio di firma con ML-DSA:
MLDsaAlgorithm alg = MLDsaAlgorithm.MLDsa65;

using (MLDsa key = MLDsa.GenerateKey(alg))
{
    byte[] data = "Messaggio da firmare"u8.ToArray();
    byte[] signature = new byte[alg.SignatureSizeInBytes];

    key.SignData(data, signature);
    bool verified = key.VerifyData(data, signature);

    Console.WriteLine($"Firma verificata: {verified}");
}

Entrambe le famiglie di algoritmi sono disponibili in più varianti — MLKem512, MLKem768, MLKem1024 per il key encapsulation, e MLDsa44, MLDsa65, MLDsa87 per la firma — con un compromesso crescente tra sicurezza e dimensione delle chiavi. Ad esempio, per MLDsa65 la chiave pubblica occupa 1952 byte, quella privata 4032 byte e la firma 3309 byte: numeri sensibilmente più grandi rispetto a ECDSA, un dettaglio da tenere presente quando si progettano protocolli o formati di messaggio con vincoli di banda o storage.

Un paio di note pratiche per chi vuole sperimentare da subito: su Linux serve OpenSSL 3.5 o superiore, mentre chi è ancora su .NET Standard 2.0 può accedere alle stesse API tramite il pacchetto NuGet Microsoft.Bcl.Cryptography. Per il supporto lato TLS 1.3, certificati basati su ML-DSA e SLH-DSA funzionano già in scenari di autenticazione quando sia il sistema operativo sia la controparte della connessione supportano i nuovi algoritmi — un altro motivo per cui la coordinazione tra client e server nella transizione non è opzionale.

Da dove cominciare, in pratica


Per un team che affronta questo tema per la prima volta, un percorso ragionevole è: primo, effettuare un inventario reale di certificati, librerie crittografiche e protocolli in uso, distinguendo per criticità e durata di vita dei dati protetti; secondo, verificare che l’infrastruttura TLS sia già su 1.3 ovunque possibile, perché è il prerequisito tecnico per tutto il resto; terzo, iniziare a sperimentare le nuove API — su .NET 10 il costo di ingresso è basso, si tratta di poche righe di codice — in ambienti non di produzione, per capire l’impatto su dimensioni di chiavi, firme e tempi di calcolo prima che diventi un requisito con scadenza stretta.

La crittografia post-quantistica non è ancora un’emergenza operativa per la maggior parte delle organizzazioni, ma il messaggio di Microsoft è chiaro: il momento di mappare le proprie dipendenze crittografiche e cominciare a testare gli algoritmi è adesso, non quando arriverà l’obbligo normativo o il primo incidente.

Fonte originale: Microsoft Urges Threat Modeling to Prepare for Post-Quantum Cryptography Migration, Petri IT Knowledgebase


The Pirate Post ha ricondiviso questo.

The media in this post is not displayed to visitors. To view it, please go to the original post.

The media in this post is not displayed to visitors. To view it, please go to the original post.

✨ DragonDoll: lo spyware Android che si finge un aggiornamento Chrome e legge Telegram, WhatsApp e Signal in 26 Paesi
#CyberSecurity
insicurezzadigitale.com/dragon…

@informatica


DragonDoll: lo spyware Android che si finge un aggiornamento Chrome e legge Telegram, WhatsApp e Signal in 26 Paesi


Si parla di:
Toggle

Una pagina che imita alla perfezione la schermata di aggiornamento di Google Chrome, un dropper che scarica il payload reale solo dopo, e un abuso sistematico dei servizi di accessibilità Android per leggere in tempo reale le chat di Telegram, WhatsApp e Signal senza mai toccare la loro crittografia end-to-end. È DragonDoll, lo spyware commerciale che il Positive Technologies Expert Security Center (PT ESC) ha tracciato in almeno 26 Paesi, Russia inclusa, con circa 150 campioni caricati in appena due mesi su un repository GitHub pubblico.

Un finto aggiornamento, un dropper vero


La catena d’infezione comincia con un APK che si spaccia per “Chrome.apk” e adotta la tecnica dei tampered headers per ostacolare gli strumenti di analisi statica. Una volta installato, il file mostra una pagina-esca che riproduce fedelmente il branding di un aggiornamento ufficiale di Google Chrome, convincendo l’utente a proseguire con l’installazione. Solo a questo punto il dropper scarica ed esegue il payload reale, denominato internamente K3iwv7VF.apk, decifrato con AES-256-GCM tramite implementazione nativa (libreria compilata in C/C++, non in Java/Kotlin), un’ulteriore misura anti-analisi rispetto ai più comuni spyware Android scritti interamente in linguaggio gestito.

Il vero motore: l’abuso dell’AccessibilityService


DragonDoll non sfrutta exploit del kernel o vulnerabilità 0-day: si affida quasi interamente all’AccessibilityService di Android, l’API pensata per rendere il sistema operativo utilizzabile da persone con disabilità visive. Una volta ottenuto il permesso — spesso concesso dalla vittima stessa, ingannata dalla schermata di richiesta mascherata da parte del sistema — lo spyware può:

  • Intercettare tap, gesture e input da tastiera in tempo reale su qualsiasi app in primo piano.
  • Leggere il contenuto delle notifiche, incluse quelle di Telegram (liste chat, contatti, testo dei messaggi e timestamp) man mano che arrivano, senza dover violare la cifratura del protocollo.
  • Monitorare visivamente lo schermo durante l’uso di WhatsApp e Signal, catturando ciò che compare a video anche quando l’app stessa è protetta da crittografia end-to-end.
  • Iniettare overlay fasulli sopra app bancarie per sottrarre credenziali (banking overlay injection), una tecnica presa in prestito dai trojan bancari Android più maturi.
  • Raccogliere contatti, SMS, log delle chiamate, screenshot, stato di batteria/Wi-Fi/USB/SIM, IMEI e la lista completa delle app installate.

Per app meno “prioritarie” come Viber, il malware si limita a una raccolta più semplice dei dati visibili a schermo, segno di uno sviluppo modulare e mirato piuttosto che di un tool generico riadattato.

Una campagna globale, con targeting linguistico chirurgico


PT ESC ha individuato la campagna per la prima volta nella primavera 2026, durante l’analisi di attacchi contro utenti in Arabia Saudita, per poi scoprire un’infrastruttura di distribuzione molto più ampia: oltre 26 Paesi colpiti, tra cui Russia, Cina, Corea del Sud e diversi Stati dell’area MENA, con il malware localizzato in 34 lingue di interfaccia. Tra il 6 marzo e il 6 maggio 2026 sono stati caricati circa 150 campioni unici su un account GitHub (kesmanta24, registrato con l’indirizzo kesmantes52@outlook.com), arrivati almeno alle versioni 9.3 e 9.4 — un ritmo di sviluppo che tradisce un progetto attivamente mantenuto e monetizzato, non un esperimento estemporaneo.

L’infrastruttura di comando e controllo si appoggia al dominio channelzones[.]co, ospitato su hosting russo, con canali di comunicazione bidirezionale basati su Socket.IO e autenticazione tramite OkHttp3; la cifratura del traffico combina AES-256-CBC e RSA OAEP in uno schema ibrido. La distribuzione iniziale avviene tramite domini civetta come datewithmealways[.]site, datewithmealways[.]online e digitaladstracking[.]com, pensati per superare filtri di reputazione e ingannare l’utente con nomi apparentemente innocui o a tema dating/advertising.

Perché conta: il mercato in espansione dello spyware commerciale Android


DragonDoll si inserisce in un trend che i ricercatori osservano da tempo: la proliferazione di famiglie spyware Android vendute o distribuite su scala industriale, capaci di aggirare la crittografia end-to-end di app come Signal e WhatsApp non attaccando il protocollo, ma il dispositivo stesso attraverso funzioni di sistema legittime come l’accessibilità. È lo stesso principio dietro altre famiglie recenti individuate mascherate da versioni “clonate” di Telegram e Signal distribuite anche tramite store ufficiali, un problema che ha spinto più volte Google a rafforzare i controlli su Play Protect senza però eliminare il rischio quando l’installazione avviene tramite sideloading da pagine web esterne, come nel caso di DragonDoll.

La presenza della Russia tra i Paesi bersaglio, insieme a un’infrastruttura C2 ospitata proprio su hosting russo, è un dettaglio che PT ESC segnala senza sciogliere del tutto l’ambiguità sull’attribuzione: non è chiaro se si tratti di un operatore locale che colpisce anche il proprio mercato interno, di un tool venduto a più clienti con targeting indipendente, o di un’operazione di sorveglianza mirata mascherata da campagna commerciale opportunistica.

Come proteggersi


  • Non installare mai aggiornamenti di Chrome (o di qualsiasi altra app) da pagine web esterne al Google Play Store: gli aggiornamenti legittimi non vengono mai proposti così.
  • Verificare periodicamente in Impostazioni → Accessibilità quali app hanno permessi attivi e revocarli per tutto ciò che non sia uno strumento di assistenza riconosciuto.
  • Diffidare di APK scaricati da domini a tema dating, advertising o “aggiornamento urgente”, tipici vettori di distribuzione di questa famiglia.
  • Su dispositivi aziendali o ad alto rischio, disabilitare il sideloading e imporre policy MDM che blocchino l’installazione da fonti sconosciute.
  • Monitorare traffico di rete verso i domini e l’infrastruttura elencati di seguito come indicatori di compromissione.


Indicatori di compromissione

# Infrastruttura C2 e distribuzione
channelzones[.]co                 (C2 primario, hosting russo, canale Socket.IO)
datewithmealways[.]site           (dominio di distribuzione)
datewithmealways[.]online         (dominio di distribuzione)
digitaladstracking[.]com          (dominio di distribuzione)

# Account di sviluppo
GitHub user: kesmanta24
Email associata: kesmantes52@outlook.com
Periodo attivita': 6 marzo - 6 maggio 2026 (~150 campioni, versioni 9.3-9.4)

# Nomi file osservati
Chrome.apk        (dropper iniziale, tampered headers)
K3iwv7VF.apk       (payload finale, AES-256-GCM nativo)

# Paesi colpiti (parziale)
Arabia Saudita, Russia, Cina, Corea del Sud, area MENA (26+ Paesi totali)
Localizzazione: 34 lingue di interfaccia

# Cifratura
Dropper -> payload: AES-256-GCM (nativo)
C2: AES-256-CBC + RSA OAEP (ibrido), autenticazione OkHttp3, canale Socket.IO

Fonte primaria: Positive Technologies Expert Security Center (PT ESC). Dettagli tecnici e IoC aggiornati al momento della pubblicazione; per l’elenco completo di hash e domini si rimanda al report integrale dei ricercatori.

The Pirate Post ha ricondiviso questo.

Wer ein Apple-Gerät hatte, wurde in der Vergangenheit recht gut vor dem Zugriff der Werbebranche gewarnt. Nach einem auch von deutschen Zeitungsverlagen angestrengten Verfahren beim Kartellamt wird der Schutz nun verwässert – mit klaren Nachteilen für die Nutzer:innen. Ein Kommentar.

netzpolitik.org/2026/absurde-e…

in reply to netzpolitik.org

Auf Kartellbehörden oder die EU braucht niemand zu warten oder zu hoffen.

Mit dieser Erkenntnis gilt: Digitale Selbstverteidigung ist der einzige Weg aus dem Werbe-Irrsinn. Politik und Recht schützen uns nicht, also unterlauft, boykottiert, behindert und blockiert die Täter selber.

Es gibt viele Mittel und Wege, und sachkundige Menschen, die Dir dabei helfen. Aber wir kommen nicht darum, uns selbst Sachkunde zu erarbeiten, Hand anzulegen und uns gegen den Überwachungskapitalismus zu wehren. People before Business.

Big Tech boykottieren, und Menschen unterstützen, die Hilfe dabei brauchen. Wir müssen lernen, solidarische Netzbürger zu werden, und uns unsere digitale Freiheit zurückzuholen. Anderenfalls werden wir von den Konzernen eingeseift und scharf rasiert.

Questa voce è stata modificata (3 settimane fa)

Perché i Pirati vogliono una riforma del diritto d’autore


Di seguito il post pubblicato oggi dal Partito pirata europeo Qualcuno una volta ha passato una serata a scegliere l’ordine di dodici canzoni, ha scritto a mano i titoli sul foglietto illustrativo e ha dato la cassetta a una persona che sperava di impressionare. Era una copia. Era stata fatta senza chiedere il permesso a nessuno. In gran parte d’Europa, inoltre, rientrava a pieno titolo in un’…

Source

reshared this

The Pirate Post ha ricondiviso questo.

Perché i Pirati vogliono una riforma del diritto d'autore

Dal dibattito sulle cassette al digitale: quando i lucchetti tecnologici e i filtri di rete scavalcano i diritti d'uso e la libertà degli utenti diventa obbligatorio reagire

pirati.io/2026/08/perche-i-pir…

@pirati@feddit.it

The Pirate Post ha ricondiviso questo.

The media in this post is not displayed to visitors. To view it, please go to the original post.

🕶️ Metas KI-Brillen stehen wegen weitreichender Datenschutzbedenken immer öfter in der Öffentlichkeit. 📺 noyb-Datenschutzjurist Felix Mikolasch hat mit dem ORF über die Risiken und Gefahren gesprochen:

👉 orf.at/av/video/onDemandVideoN…

Questa voce è stata modificata (4 settimane fa)
The Pirate Post ha ricondiviso questo.

Meine Atmung fühlte sich beim Lesen eingeengt an, vlt. weil ich zwischendrin die Luft angehalten habe.


Würde die AfD ein Bundesland regieren, hätte sie eine Reihe von Möglichkeiten, den Sicherheitsapparat in ihrem Sinne neu auszurichten. Doch Demokrat*innen stehen dem nicht hilflos gegenüber. Wir haben mit dem Rechtswissenschaftler Markus Thiel darüber gesprochen, welche Gefahren eine potenzielle Machtübernahme birgt und was man dagegen tun kann. netzpolitik.org/2026/polizei-u…

The Pirate Post ha ricondiviso questo.

Für alle, die noch im Ferienmodus sind: Es ist wieder #DNIPBriefing-Dienstag! (Und auch für alle anderen, BTW)

1️⃣ Um Video-Überwachung (und ob man sich mit entsprechender Kleidung dagegen schützen kann) ging es schon im Briefing von letzter Woche. Dabei kamen wir zum Schluss, dass man sich nicht auf diesen Schutz verlassen sollte, da keine systematischen Tests bekannt sind und es viele unterschiedliche Überwachungs-Technologie mit unterschiedlichen Eigenschaften gibt.

dnip.ch/2026/08/18/dnip-briefi…

in reply to Marcel Waldvogel

Und wer die automatische Erkennung von potentiell strafrechtlich relevanten Situationen für Zukunftsmusik hält: Wie netzpolitik.org letzte Woche offengelegt hat, arbeitet zumindest die Mannheimer Polizei seit mehreren Jahren daran, Gefahrensituationen oder unerwünschtes oder gefährliches Verhalten durch Software zuverlässig zu erkennen.
netzpolitik.org/2026/verhalten…

reshared this

in reply to Marcel Waldvogel

Dazu wurden mit Probanden bisher rund 2’000 Kurzfilme mit „auffälligem“ Verhalten erstellt, welche für die Erkennungs-Software als Test- und Trainingsmaterial genutzt werden. Da dies offenbar nicht reicht, soll die Menge der Filme auf bis zu 10’000 ausgebaut werden. Handfeste Belege über Wirksamkeit, Fehlerquoten und Verlässlichkeit der damit getesteten/trainierten Software oder gar unabhängige Evaluationen gibt es bisher allerdings nicht.

reshared this

in reply to Marcel Waldvogel

2️⃣ Überwachung wird ja gerade „demokratisiert“, wie man euphemistisch sagen könnte. Für wenige hundert Franken kann jeder zum Datensammel-Agenten von Meta werden. Die Digitale Gesellschaft sieht darin einen „Frontalangriff auf unsere Privatsphäre“ und verlangt ein grundsätzliches Verbot solcher anlassloser Überwachung durch Smart Glasses mit streng definierten Ausnahmen, die sich strikt an „Privacy by Design“ orientieren müssten.
digitale-gesellschaft.ch/2026/…

reshared this

in reply to Marcel Waldvogel

In Deutschland läuft eine Strafanzeige von HateAid gegen Meta, den Brillenhersteller sowie Optik-Handelsketten. Nach Ansicht von HateAid verstösst der Verkauf gegen das deutsche Telekommunikation-Digitale-Dienste-Datenschutz-Gesetz (TDDDG), das Geräte verbietet, welche als Alltagsgegenstände getarnt seien, aber dazu bestimmt sind, andere unbemerkt zu filmen oder abzuhören.
heise.de/news/KI-Brillen-Straf…

reshared this

in reply to Marcel Waldvogel

Das Verfahren dürfte spannend werden, da es in Deutschland entsprechende Präzedenzfälle gibt: So wurde 2017 eine „Smarte Puppe“ verboten, da sie eine „verbotene Sendeanlage“ darstelle und damit einem Spionagegerät gleichgestellt sei. Sie wurde vom Markt genommen und Eltern aufgefordert, die Puppen zu vernichten, denn alleine der Besitz eines solchen Spionagegeräts ist strafbar.
netzpolitik.org/2017/schnueffe…

reshared this

in reply to Marcel Waldvogel

Wer sich in der Schweiz schon einmal betätigen will: Campax hat eine Petition gestartet, die ein Verbot von Smart Glasses fordert und der man sich anschliessen kann. Dort wird insbesondere das heimliche Filmen von Frauen und Kindern mit anschliessender Veröffentlichung ohne Zustimmung der Betroffenen thematisiert, beispielsweise in Badis und Umkleidekabinen.
#SmartGlasses #Überwachung #Meta
act.campax.org/petitions/smart…

reshared this

in reply to Marcel Waldvogel

(Ach ja: Bereitet euch schon einmal geistig auf Ohrstöpsel mit Kameras vor. Die sind auch allgegenwärtig in Gyms und Umkleidekabinen.)
macrumors.com/2026/08/17/camer…

reshared this

in reply to Marcel Waldvogel

3️⃣ Wie The Verge schreibt, hat Amazon einen neuen Weg gefunden, Kunden auf die eigene Webseite zu locken: In den Bestätigungsmails zu Bestellungen und deren Versandstatus wird nicht mehr das spezifische Produkt („20 Stück Kernseife“) erwähnt, sondern nur noch die Produkt-Kategorie („Hygenieprodukt“). Wer viel bestellt und genau wissen will, das denn da nun unterwegs ist, muss den Bestell-Status direkt auf der Amazon-Webseite nachschlagen.
theverge.com/ai-artificial-int…

reshared this

in reply to Marcel Waldvogel

Offiziell begründet Amazon dies damit, dass so die Weitergabe von Kundeninformationen ausserhalb der Amazon-App und -Website reduziert wird, was die Privatsphäre der Kunden besser schützt.
dnip.ch/2026/08/18/dnip-briefi…

reshared this

in reply to Marcel Waldvogel

The Verge spekuliert im Artikel darüber, dass der wahre Grund wohl eher darin liegt, dass Google (via Gmail) die Inboxes ihrer Nutzenden scannt und die dabei gefundenen Produktdetails für Profiling und gezielte Werbung nutzen kann. Profiling und Werbung, welche dazu führen können, dass Kunden halt auch Angebote ausserhalb des Amazon-Universums kennenlernen …
dnip.ch/2024/03/26/der-taeglic…

reshared this

in reply to Marcel Waldvogel

4️⃣ Vor einigen Wochen berichteten wir, dass in letzter Zeit vermehrt grosse Bestellungen bei Antiquariaten eingingen. Vermutet wurde, dass die KI-Konzerne im grossen Stil Bücher aufkaufen würden, um ihre Inhalte dann zu scannen. Der Einfachheit halber (und dem Urheberrecht zuliebe) würden die Bücher dabei vernichtet und entsorgt, so die Vermutung. Genaueres wusste man bisher aber nicht, da die Bestellungen über Zwischenhändler und Dienstleister liefen.
dnip.ch/2026/06/23/dnip-briefi…

reshared this

in reply to Marcel Waldvogel

Diese Vermutung wurde inzwischen bestätigt. 404 Media hat einer dieser Bücher-Grosslieferungen eine Apple AirTag untergejubelt, mit dem sie die Buchlieferung quer durch die USA und zurück verfolgen konnten (Paywall). Die Lieferung landete ironischerweise im gleichen Amazon-Gebäudekomplex in Las Vegas, in dem auch Bücher On-Demand gedruckt werden. Nur, dass mit diesen angelieferten Büchern so ziemlich genau das Umgekehrte passiert.
404media.co/we-tracked-a-shipm…


We Tracked a Shipment of Rare Books. It Ended at an Amazon AI Training Facility


Amazon is buying massive quantities of books, scanning them for AI training data, and destroying them in the process.

A 404 Media investigation was able to reveal Amazon’s book buying operation, which hasn’t been previously reported, by placing a tracking device in a rare book we suspected would be acquired by an AI company for training data, and following it around the country to its final destination.

📖
Do you know work at a facility where you scan books? I would love to hear from you. Using a non-work device, you can message me securely on Signal at @emanuel.404. Otherwise, send me an email at emanuel@404media.co.

That final destination was an Amazon warehouse in Las Vegas, Nevada. Amazon employees who work at this location say all they do is receive massive shipments of printed books which they then cut the bindings off in order to scan the books more quickly. The printed book is destroyed in the process. The logo of the Amazon team that works at this warehouse, called VGT3, is a dinosaur, brandishing its teeth and with a book in its hands.

This post is for subscribers only


Become a member to get access to all content
Subscribe now


reshared this

in reply to Marcel Waldvogel

In internen Foren wurde die Arbeit als für Amazon-Verhältnisse vergleichsweise stressfrei beschrieben, wie 404 Media berichtet. Die Mitarbeitenden dort würden die Bücher aufschneiden und zuerst die ISBN-Barcodes, später die Seiteninhalte scannen.

Im Artikel wird vermutet, dass vor allem Bücher vor 2022 aufgekauft würden, da diese garantiert ohne LLM generiert worden seien.

Ob Amazon auch Bücher aufkauft, die zuvor im gleichen Gebäude gedruckt wurden, ist nicht bekannt.
dnip.ch/2026/08/18/dnip-briefi…

reshared this

in reply to Marcel Waldvogel

5️⃣ Apple ist bekannt dafür, dass sie Privatsphäre ernster nehmen als andere IT-Grosskonzerne. So haben sie früh begonnen, die Trackingmöglichkeiten durch Apps einzuschränken bzw. die Datenverwendung durch App-Anbieter zu dokumentieren. Apple bietet seit mehreren Jahren die Möglichkeit an, den Apps das Datensammeln mit wenigen Klicks zu untersagen. Nun hat allerdings das deutsche Bundeskartellamt geurteilt, dass das nicht mehr so einfach gehen dürfe.
netzpolitik.org/2026/bundeskar…

reshared this

in reply to Marcel Waldvogel

Geklagt hatten deutsche Werbe- und Zeitungsverbände. Einer der Klagepunkte betrifft die Ungleichbehandlung von Apple und Drittparteien in Apples App Tracking Transparency Framework (ATTF), der definitiv in die Zuständigkeit einer Wettbewerbsbehörde fällt.

Das Bundeskartellamt hat aber auch entschieden, dass die App-Tracking-Anbieter neu 4000 Zeichen Platz im Tracking-Ja/Nein-Dialog bekommen sollen, um die Vorteile von personalisierter Werbung und Tracking zu erklären.
dnip.ch/2026/08/18/dnip-briefi…

reshared this

in reply to Marcel Waldvogel

Das Wort „Tracking“ darf Apple übrigens in dem Zusammenhang nicht mehr verwenden.

Dabei dürfte es zumindest unseren Leser:innen durchaus geläufig sein, dass die Datensammelei mit Datenschutzrisiken verbunden ist. Die Daten bleiben nämlich nicht bei den App-Anbietern, die Interessen und Standortdaten der Nutzer:innen werden fleissig und oft unkontrolliert erfasst und weitergegeben.
dnip.ch/2025/12/16/dnip-briefi…

reshared this

in reply to Marcel Waldvogel

Das Wort „Tracking“ darf Apple übrigens in dem Zusammenhang nicht mehr verwenden.

Dabei dürfte es zumindest unseren Leser:innen durchaus geläufig sein, dass die Datensammelei mit Datenschutzrisiken verbunden ist. Die Daten bleiben nämlich nicht bei den App-Anbietern, die Interessen und Standortdaten der Nutzer:innen werden fleissig und oft unkontrolliert erfasst und weitergegeben.

Questa voce è stata modificata (4 settimane fa)
in reply to Marcel Waldvogel

Wer kennt nicht die Dialoge, dass dem Anbieter „die Privatsphäre wichtig“ sei und die Daten nur mit ausgewählten „949 Partnern“ geteilt würden? Dass einige dieser Partner die Daten weiter teilen und damit auch sicherheitsrelevante Informationen geleckt werden, wird damit vermutlich zukünftig unterschlagen werden müssen.
dnip.ch/2025/12/16/dnip-briefi…

reshared this

in reply to Marcel Waldvogel

6️⃣ Anthropic hat angekündigt, in von Claude generierten Texten zukünftig Wasserzeichen einzufügen, also für die Leserin nicht erkennbare Indikatoren, dass ein Text von Claude erstellt wurde. Wasserzeichen kennen wir ja primär von Bildern und im ersten Moment scheint es fraglich, wie sowas in Texten (die ja schlussendlich simple Zeichenketten sind) funktionieren soll.
anthropic.com/news/claude-text…

reshared this

in reply to Marcel Waldvogel

Weiterhelfen kann hierbei dieser Text⬇️, der den Effekt auch gleich visuell aufbereitet. Kurz gesagt wird ausgenutzt, dass bei LLM-gestützter Textgenerierung das nächste Wort durch das LLM wahrscheinlichkeitsbasiert aus einer Menge von möglichen Wörtern ausgewählt wird. Oft handelt es sich dabei um Synonyme, und man kann mit einer Änderung der Gewichtung erreichen, dass gewisse Synonyme häufiger ausgewählt werden als dies rein basierend auf den Trainingsdaten gegeben wäre.
declaude.org/watermarking/

reshared this

in reply to Marcel Waldvogel

Dies bedeutet konkret, dass nur Anthropic erkennen kann, ob ein Text durch Claude erstellt oder signifikant bearbeitet wurde. Nur Anthropic kennt schliesslich die geänderte Gewichtung und kann so durch eine Textanalyse überprüfen, ob in einem vorliegenden Text dieser Gewichtung entspricht. Allerdings sind hierzu längere Texte (typischerweise mehrere Seiten) notwendig, da sich in kurzen Texten kaum genügend Synonyme austauschen lassen können.
dnip.ch/2023/01/30/wie-funktio…

reshared this

in reply to Marcel Waldvogel

Zu Kontroversen führt momentan, inwieweit durch das Watermarking die Bedeutung von Texten verändert wird und inwieweit es statthaft ist, Wasserzeichen auch dann anzubringen, wenn Claude nur zum Redigieren/Korrigieren von durch Menschen geschriebenen Texten verwendet wird. Wir werden darauf in einem separaten Artikel noch eingehen.

reshared this

in reply to Marcel Waldvogel

7️⃣ Klaus Landefeld, der u.a. langjährig für DE-CIX tätig war, die in Frankfurt einen der grössten Internetknoten der Welt betreiben, kritisierte im Heise-Interview die Wünsche der Ermittlungsbehörden nach mehr Vorratsdatenspeicherung mit den Worten: "Wenn man aber bei einer Anzeige noch nicht mal die Daten abfragt, die der Telekommunikationsprovider extra bereitgestellt hat, ist die ganze Argumentation mit zu kurzen Speicherfristen etwas schief."
heise.de/hintergrund/Missing-L…

reshared this

in reply to Marcel Waldvogel

Überwachung und Vorratsdatenspeicherung sind Themen, in dem auch aktuell in der Schweiz die Digitale Gesellschaft @digiges wieder aktiv ist.

#Überwachung #Vorratsdatenspeicherung #QuickFreeze
digitale-gesellschaft.ch/quick…

reshared this

in reply to Marcel Waldvogel

8️⃣ Die Weko ist fleissig. Nicht nur ist sie in (Vor-)Abklärungen mit Microsoft (Preiserhöhungen) und Apple (NFC), neu wird auch gegen Google ermittelt. Grund: in der Schweiz kann im Gegensatz zu den Nachbarländern die Suchmaschine auf Android nicht mehr beim Start ausgewählt werden, sondern ist initial fix auf die Google-eigene festgelegt.
weko.admin.ch/de/newnsb/a1GWSd…

reshared this

in reply to Marcel Waldvogel

PS: In einem Kommentar auf unserer Seite hat Rolf darauf hingewiesen, dass es auch andere Tracking-Methoden gibt. Im Funkbereich reichen bereits ein paar "normale" WLAN-Accesspoints, um Menschen zu lokalisieren und sie wiederzuerkennen.
heise.de/hintergrund/Menschen-…

reshared this

Why Pirates want Copyright Reform


The media in this post is not displayed to visitors. To view it, please log in.

Someone once spent an evening choosing the order of twelve songs, wrote the titles on the paper insert by hand, and gave the cassette to a person they were hoping to impress. It was a copy. It was made without asking anyone. Across much of Europe it also fell squarely within an exception that lawmakers had gone out of their way to create.

That was not an oversight. Germany wrote a private copying exception into its Copyright Act in 1965 and paid for it with a levy on recording equipment and blank media. Most of continental Europe followed the same model, and the arrangement was eventually carried into EU law as the private copying exception in Article 5(2)(b) of the InfoSoc Directive — the 2001 law that still governs copyright across the European Union. Rightsholders were compensated. Nobody had to be watched.

The record industry campaigned against it anyway. “Home taping is killing music” was the British Phonographic Industry’s slogan from 1981, printed on record sleeves under a skull-and-crossbones. It is worth noticing what that campaign did with language: an act that European lawmakers had deliberately permitted, and for which artists were already being paid through the levy, was described as killing. The framing arrived two decades before file-sharing became a mass phenomenon — the first clue that the argument was never really about the copying.

Two answers, and the one that travelled


Europe and the United States faced the same question in the cassette era and answered it in opposite ways.

Europe legalised the copying and paid for it collectively. Levies on recording equipment and blank media flow through collecting societies to authors, performers, publishers and visual artists. According to the global study published by the collecting societies themselves, the system now operates in 31 European countries against four in the whole of the Americas, generates over €1 billion a year worldwide, and costs each consumer between €1.14 and €3.19 annually. In more than twenty countries part of the money funds social and cultural programmes — training, pensions, support for emerging artists — which is why this is better understood as social infrastructure than as a tax. The Court of Justice has held that up to half of such compensation may be directed to those institutions. And the study describes the defining feature of the whole arrangement in a single phrase: the money reaches creators “without monitoring individual users.”

The system has real critics. Europe’s technology industry argues that levies are an outdated instrument that charges buyers for copies many of them never make, applied through tariffs that vary widely and unpredictably between Member States. That is a fair objection to how the levy is calculated. It is not an argument against the principle underneath it.

The United States built nothing comparable. Its Audio Home Recording Act of 1992 did create a royalty, but only on dedicated digital audio recording devices — and when the recording industry tried to apply it to the first MP3 players, the Ninth Circuit held in RIAA v. Diamond Multimedia (1999) that the Rio was not such a device, because it merely made portable what was already on a computer’s hard drive. General-purpose computers fell outside the scheme, and the scheme never recovered. One comparison put annual private copying payments at $1.2 billion in the EU against roughly $1 million in the United States.

With no mechanism to compensate for copying, there was nothing to offer in exchange for permitting it. The American answer to home copying could only be to prevent it. That answer was written into the WIPO Copyright Treaty in 1996, into the Digital Millennium Copyright Act in 1998, and then into European law in the InfoSoc Directive of 2001 — the very directive that codified the private copying exception. Europe kept its levy and imported the lock in the same instrument.

Europeans consequently pay twice. The levy is charged on the device on the assumption that private copies will be made. The DRM on the file then prevents the copy the levy has already paid for.

Canada shows where that combination ends. It adopted the European model in 1997, with a private copying exception for music funded by a levy — but the statute named media rather than devices. When the Copyright Board extended the tariff to the memory inside MP3 players, the Federal Court of Appeal struck it down for want of authority, and the Supreme Court declined to intervene. The levy stayed on blank CDs while the copying moved to phones, and receipts fell from C$38 million in 2004 to C$1.1 million in 2019. Canada then implemented the anti-circumvention rules in 2012 regardless, alongside a set of user exceptions more generous than anything in EU law — format-shifting, backup, and a non-commercial remix exception Europe still lacks — all of which the digital locks override. The lesson is worth stating precisely: the Canadian levy did not fail because collective compensation does not work. It failed because a statute written around cassettes was never rewritten.

The companies that came to operate the distribution layer grew up under the American rule set rather than the European one. Apple’s iTunes Store, Amazon’s Kindle, Google’s YouTube, Netflix and Steam determine the terms on which most Europeans now reach music, books, film and games, with Spotify the notable European exception. The safe harbour that made platforms built on user uploads commercially viable was an American design. Europe’s eventual answer, Article 17, imposes filtering costs — Google says it has invested over $100 million in Content ID alone — that in practice only the largest incumbents can absorb.

Meanwhile the European mechanism erodes. Levy collections across 32 countries have been broadly flat in cash terms since 2017 and have fallen in real terms, because the copying has moved to streaming, where the exception, and the payment attached to it, does not clearly reach.

The copy you cannot give away


Buy a paperback and it is yours. Resell it, lend it, give it away, leave it to someone in a will — copyright does not stand in the way, because the right to control distribution of that copy is exhausted by the first authorised sale. This is not a loophole. Exhaustion is the mechanism that makes libraries, second-hand shops and gifts possible.

Buy the same book as an e-book and none of it holds. In Tom Kabinet (C-263/18, 19 December 2019), the Court of Justice of the European Union ruled that supplying an e-book for permanent download is not “distribution” at all but “communication to the public” — a right that is never exhausted. A second-hand market in e-books is therefore unlawful without the publisher’s permission. The Court had reached the opposite conclusion for downloaded software seven years earlier, in UsedSoft (C-128/11), which tells us the distinction is a policy choice rather than a fact of nature.

No piracy is involved anywhere in this. A reader who paid full price cannot pass the book on. What was removed was not an unauthorised act but an ordinary one.

The lock that outranks the right


The private copying exception still exists on paper. Exercising it is another matter.

The anti-circumvention rules imported in 2001 make it unlawful to defeat a technological protection measure — the family of access controls usually called DRM. Article 6(4) was meant to be the safety valve between the two halves of the directive: Member States should ensure that people can still benefit from certain exceptions even when a work is locked. In practice it applies to a closed list of exceptions, leaves the remedy to national procedures that most Member States never built, and offers nothing at all where a work is streamed on demand.

The result is a right that cannot be exercised. The lock wins, and the lock does not distinguish between an infringing copy and a lawful one — it cannot, because it does not know what the user intends.

Portugal is the exception that proves the point. Law 36/2017 permits circumvention where the purpose is to exercise a lawful exception, and forbids DRM on public domain works and on works funded by public money. Campaigners there needed fifteen years to win a right that the Directive had always claimed to guarantee.

Enforcement moved into the pipes


Article 17 of the 2019 Copyright in the Digital Single Market Directive made large platforms liable for what their users upload unless they make best efforts to prevent it. It was argued for as a fix to the “value gap” between what platforms earn and what rightsholders receive. What it requires in practice is automated inspection of everything anyone posts.

The Court of Justice upheld the provision in Poland v Parliament and Council (C-401/19, 26 April 2022), but not in the form several governments wanted. It rejected the argument that content could be blocked systematically at upload so long as users could appeal afterwards, and held that safeguards must operate before blocking, not after. A filter that cannot tell a quotation from an infringement may not be deployed as though it can.

That judgment is a genuine win for user rights. It is also an admission about the mechanism: the default behaviour of automated enforcement is to remove lawful speech, and it took the EU’s highest court to say so.

When there is no copy at all


The clearest case is the one currently on the table.

The European citizens’ initiative “Stop Destroying Videogames” was submitted to the Commission on 26 January 2026with 1,294,188 verified signatures. It asked for something modest: that a game sold to the public should be left in a playable state when the publisher stops supporting it. On 16 June 2026 the Commission replied that it cannot at this stage propose such an obligation, and offered instead to convene the industry on a voluntary code of conduct by the end of 2026.

The background to that request is Ubisoft’s racing game The Crew, delisted in December 2023 and switched off three months later. Buyers then found their licences revoked outright. In the litigation that followed — including a case brought by the French consumer association UFC-Que Choisir — Ubisoft’s defence was that customers had never bought a copy at all, only a limited licence to access a service.

That defence is legally coherent, and that is precisely the problem. When the transaction is a licence rather than a copy, there is nothing to lend, resell, archive or hand to anyone. The unmonitored exchange of culture between people does not have to be prohibited. It simply has nothing left to move.

The case on the other side


It deserves stating properly. Creators need income, and an unlimited right to copy and redistribute commercially would remove the market that pays them. Rightsholders cannot realistically police millions of individual users, so building enforcement into infrastructure is a rational response to a real problem. Licensed streaming has been a genuine success on its own terms: cheap, legal and convenient, and its defenders can fairly point out that it did more to draw people away from unauthorised downloading than a decade of lawsuits ever managed.

None of that is in dispute. The question is what the mechanisms actually deliver.

Extending the term of protection for sound recordings from 50 to 70 years, as the EU did in 2011, prevents no unauthorised copying whatsoever — it can only extend control over recordings that already exist. A study prepared for the European Parliament sets out the recurring criticism: the benefit falls mainly to labels and a small number of successful performers, while ordinary session musicians have little bargaining power over the contracts that determine what reaches them. The levy that paid for the mixtape paid artists directly. Digital rights management, upload filters and licence-only distribution do not pay them more; they change who decides on what terms a work reaches the public.

That is the reform question. Not whether creators should be paid — they should, and better than they currently are — but who holds the switch.

What reform actually asks for


The European Pirates’ programme states the position directly:

  • Non-commercial use is protected, not merely tolerated. “Copying, storing, using, and providing access to literary and artistic works for non-commercial purposes must not only be legalized but protected by law.” Non-commercial file sharing should be allowed.
  • Shorter commercial terms. The term of the commercial monopoly should be further shortened, exclusive rights must be limited in time and scope, and neither may be expanded retrospectively.
  • Derivative and everyday uses carved out. Remixing, parody, quotation and sampling exempt from the commercial monopoly; derivative works permitted by default, with exceptions written explicitly into law; linking never an infringement.
  • Exceptions made mandatory across the EU, so that a lawful use in one Member State is not an infringement in the next.
  • Public money, public access. Cultural heritage digitised and made available free of charge; an end to the privatisation of profits from publicly funded works.
  • Transparency from collecting societies, and fair contract terms for the artists on whose behalf all of this is said to be done.

Four further steps follow from the mechanisms described above, and are where the argument now runs. A right to circumvent DRM where the purpose is a lawful one, together with a ban on DRM applied to public domain works and works funded from public money, on the Portuguese model. Digital exhaustion, so that a purchased digital copy can be resold, lent, given away and inherited like its physical equivalent. An end-of-life duty, so that a work withdrawn from sale is left usable rather than switched off — the ask that 1.29 million Europeans have already signed. And a collective compensation mechanism that survives the shift away from physical media, because the European alternative to controlling users was always to pay creators without watching them, and that half of the bargain is the half currently being allowed to lapse.

Some of this is live now. The Commission’s Digital Fairness Act, expected in the second half of 2026, is where the question of what a consumer actually owns when they buy something digital will next be argued.

The point


The mixtape was not a legal grey area that technology happened to close. It was a settled European compromise, and a deliberate one: culture could move between people without permission and without a record, and creators were paid for it through a mechanism that required watching nobody. Europe had built the thing the digital debate is still said to be searching for — a way to pay creators that does not depend on controlling what individuals do with what they own.

Then it adopted, on top of that mechanism, a legal architecture designed elsewhere for a country that had never built one. Every measure since has chipped away at one half of the European compromise while leaving the other half unimproved, and the half that erodes is the one that pays artists. Pirates want copyright reform not because copying should be free of consequence, but because a law written to reward authors has been rebuilt, piece by piece, into a system for deciding who may hand what to whom — and that is a different thing entirely.

Read more on our position, to support the review of the 2019 Copyright in the Digital Single Market Directive in the light of the even more modern advances in technology, below.

Consultation Feedback on CDSM – European PiratesDownload


europeanpirates.eu/why-pirates…


Why Pirates want Copyright Reform


Someone once spent an evening choosing the order of twelve songs, wrote the titles on the paper insert by hand, and gave the cassette to a person they were hoping to impress. It was a copy. It was made without asking anyone. Across much of Europe it also fell squarely within an exception that lawmakers had gone out of their way to create.

That was not an oversight. Germany wrote a private copying exception into its Copyright Act in 1965 and paid for it with a levy on recording equipment and blank media. Most of continental Europe followed the same model, and the arrangement was eventually carried into EU law as the private copying exception in Article 5(2)(b) of the InfoSoc Directive — the 2001 law that still governs copyright across the European Union. Rightsholders were compensated. Nobody had to be watched.

The record industry campaigned against it anyway. “Home taping is killing music” was the British Phonographic Industry’s slogan from 1981, printed on record sleeves under a skull-and-crossbones. It is worth noticing what that campaign did with language: an act that European lawmakers had deliberately permitted, and for which artists were already being paid through the levy, was described as killing. The framing arrived two decades before file-sharing became a mass phenomenon — the first clue that the argument was never really about the copying.

Two answers, and the one that travelled


Europe and the United States faced the same question in the cassette era and answered it in opposite ways.

Europe legalised the copying and paid for it collectively. Levies on recording equipment and blank media flow through collecting societies to authors, performers, publishers and visual artists. According to the global study published by the collecting societies themselves, the system now operates in 31 European countries against four in the whole of the Americas, generates over €1 billion a year worldwide, and costs each consumer between €1.14 and €3.19 annually. In more than twenty countries part of the money funds social and cultural programmes — training, pensions, support for emerging artists — which is why this is better understood as social infrastructure than as a tax. The Court of Justice has held that up to half of such compensation may be directed to those institutions. And the study describes the defining feature of the whole arrangement in a single phrase: the money reaches creators “without monitoring individual users.”

The system has real critics. Europe’s technology industry argues that levies are an outdated instrument that charges buyers for copies many of them never make, applied through tariffs that vary widely and unpredictably between Member States. That is a fair objection to how the levy is calculated. It is not an argument against the principle underneath it.

The United States built nothing comparable. Its Audio Home Recording Act of 1992 did create a royalty, but only on dedicated digital audio recording devices — and when the recording industry tried to apply it to the first MP3 players, the Ninth Circuit held in RIAA v. Diamond Multimedia (1999) that the Rio was not such a device, because it merely made portable what was already on a computer’s hard drive. General-purpose computers fell outside the scheme, and the scheme never recovered. One comparison put annual private copying payments at $1.2 billion in the EU against roughly $1 million in the United States.

With no mechanism to compensate for copying, there was nothing to offer in exchange for permitting it. The American answer to home copying could only be to prevent it. That answer was written into the WIPO Copyright Treaty in 1996, into the Digital Millennium Copyright Act in 1998, and then into European law in the InfoSoc Directive of 2001 — the very directive that codified the private copying exception. Europe kept its levy and imported the lock in the same instrument.

Europeans consequently pay twice. The levy is charged on the device on the assumption that private copies will be made. The DRM on the file then prevents the copy the levy has already paid for.

Canada shows where that combination ends. It adopted the European model in 1997, with a private copying exception for music funded by a levy — but the statute named media rather than devices. When the Copyright Board extended the tariff to the memory inside MP3 players, the Federal Court of Appeal struck it down for want of authority, and the Supreme Court declined to intervene. The levy stayed on blank CDs while the copying moved to phones, and receipts fell from C$38 million in 2004 to C$1.1 million in 2019. Canada then implemented the anti-circumvention rules in 2012 regardless, alongside a set of user exceptions more generous than anything in EU law — format-shifting, backup, and a non-commercial remix exception Europe still lacks — all of which the digital locks override. The lesson is worth stating precisely: the Canadian levy did not fail because collective compensation does not work. It failed because a statute written around cassettes was never rewritten.

The companies that came to operate the distribution layer grew up under the American rule set rather than the European one. Apple’s iTunes Store, Amazon’s Kindle, Google’s YouTube, Netflix and Steam determine the terms on which most Europeans now reach music, books, film and games, with Spotify the notable European exception. The safe harbour that made platforms built on user uploads commercially viable was an American design. Europe’s eventual answer, Article 17, imposes filtering costs — Google says it has invested over $100 million in Content ID alone — that in practice only the largest incumbents can absorb.

Meanwhile the European mechanism erodes. Levy collections across 32 countries have been broadly flat in cash terms since 2017 and have fallen in real terms, because the copying has moved to streaming, where the exception, and the payment attached to it, does not clearly reach.

The copy you cannot give away


Buy a paperback and it is yours. Resell it, lend it, give it away, leave it to someone in a will — copyright does not stand in the way, because the right to control distribution of that copy is exhausted by the first authorised sale. This is not a loophole. Exhaustion is the mechanism that makes libraries, second-hand shops and gifts possible.

Buy the same book as an e-book and none of it holds. In Tom Kabinet (C-263/18, 19 December 2019), the Court of Justice of the European Union ruled that supplying an e-book for permanent download is not “distribution” at all but “communication to the public” — a right that is never exhausted. A second-hand market in e-books is therefore unlawful without the publisher’s permission. The Court had reached the opposite conclusion for downloaded software seven years earlier, in UsedSoft (C-128/11), which tells us the distinction is a policy choice rather than a fact of nature.

No piracy is involved anywhere in this. A reader who paid full price cannot pass the book on. What was removed was not an unauthorised act but an ordinary one.

The lock that outranks the right


The private copying exception still exists on paper. Exercising it is another matter.

The anti-circumvention rules imported in 2001 make it unlawful to defeat a technological protection measure — the family of access controls usually called DRM. Article 6(4) was meant to be the safety valve between the two halves of the directive: Member States should ensure that people can still benefit from certain exceptions even when a work is locked. In practice it applies to a closed list of exceptions, leaves the remedy to national procedures that most Member States never built, and offers nothing at all where a work is streamed on demand.

The result is a right that cannot be exercised. The lock wins, and the lock does not distinguish between an infringing copy and a lawful one — it cannot, because it does not know what the user intends.

Portugal is the exception that proves the point. Law 36/2017 permits circumvention where the purpose is to exercise a lawful exception, and forbids DRM on public domain works and on works funded by public money. Campaigners there needed fifteen years to win a right that the Directive had always claimed to guarantee.

Enforcement moved into the pipes


Article 17 of the 2019 Copyright in the Digital Single Market Directive made large platforms liable for what their users upload unless they make best efforts to prevent it. It was argued for as a fix to the “value gap” between what platforms earn and what rightsholders receive. What it requires in practice is automated inspection of everything anyone posts.

The Court of Justice upheld the provision in Poland v Parliament and Council (C-401/19, 26 April 2022), but not in the form several governments wanted. It rejected the argument that content could be blocked systematically at upload so long as users could appeal afterwards, and held that safeguards must operate before blocking, not after. A filter that cannot tell a quotation from an infringement may not be deployed as though it can.

That judgment is a genuine win for user rights. It is also an admission about the mechanism: the default behaviour of automated enforcement is to remove lawful speech, and it took the EU’s highest court to say so.

When there is no copy at all


The clearest case is the one currently on the table.

The European citizens’ initiative “Stop Destroying Videogames” was submitted to the Commission on 26 January 2026with 1,294,188 verified signatures. It asked for something modest: that a game sold to the public should be left in a playable state when the publisher stops supporting it. On 16 June 2026 the Commission replied that it cannot at this stage propose such an obligation, and offered instead to convene the industry on a voluntary code of conduct by the end of 2026.

The background to that request is Ubisoft’s racing game The Crew, delisted in December 2023 and switched off three months later. Buyers then found their licences revoked outright. In the litigation that followed — including a case brought by the French consumer association UFC-Que Choisir — Ubisoft’s defence was that customers had never bought a copy at all, only a limited licence to access a service.

That defence is legally coherent, and that is precisely the problem. When the transaction is a licence rather than a copy, there is nothing to lend, resell, archive or hand to anyone. The unmonitored exchange of culture between people does not have to be prohibited. It simply has nothing left to move.

The case on the other side


It deserves stating properly. Creators need income, and an unlimited right to copy and redistribute commercially would remove the market that pays them. Rightsholders cannot realistically police millions of individual users, so building enforcement into infrastructure is a rational response to a real problem. Licensed streaming has been a genuine success on its own terms: cheap, legal and convenient, and its defenders can fairly point out that it did more to draw people away from unauthorised downloading than a decade of lawsuits ever managed.

None of that is in dispute. The question is what the mechanisms actually deliver.

Extending the term of protection for sound recordings from 50 to 70 years, as the EU did in 2011, prevents no unauthorised copying whatsoever — it can only extend control over recordings that already exist. A study prepared for the European Parliament sets out the recurring criticism: the benefit falls mainly to labels and a small number of successful performers, while ordinary session musicians have little bargaining power over the contracts that determine what reaches them. The levy that paid for the mixtape paid artists directly. Digital rights management, upload filters and licence-only distribution do not pay them more; they change who decides on what terms a work reaches the public.

That is the reform question. Not whether creators should be paid — they should, and better than they currently are — but who holds the switch.

What reform actually asks for


The European Pirates’ programme states the position directly:

  • Non-commercial use is protected, not merely tolerated. “Copying, storing, using, and providing access to literary and artistic works for non-commercial purposes must not only be legalized but protected by law.” Non-commercial file sharing should be allowed.
  • Shorter commercial terms. The term of the commercial monopoly should be further shortened, exclusive rights must be limited in time and scope, and neither may be expanded retrospectively.
  • Derivative and everyday uses carved out. Remixing, parody, quotation and sampling exempt from the commercial monopoly; derivative works permitted by default, with exceptions written explicitly into law; linking never an infringement.
  • Exceptions made mandatory across the EU, so that a lawful use in one Member State is not an infringement in the next.
  • Public money, public access. Cultural heritage digitised and made available free of charge; an end to the privatisation of profits from publicly funded works.
  • Transparency from collecting societies, and fair contract terms for the artists on whose behalf all of this is said to be done.

Four further steps follow from the mechanisms described above, and are where the argument now runs. A right to circumvent DRM where the purpose is a lawful one, together with a ban on DRM applied to public domain works and works funded from public money, on the Portuguese model. Digital exhaustion, so that a purchased digital copy can be resold, lent, given away and inherited like its physical equivalent. An end-of-life duty, so that a work withdrawn from sale is left usable rather than switched off — the ask that 1.29 million Europeans have already signed. And a collective compensation mechanism that survives the shift away from physical media, because the European alternative to controlling users was always to pay creators without watching them, and that half of the bargain is the half currently being allowed to lapse.

Some of this is live now. The Commission’s Digital Fairness Act, expected in the second half of 2026, is where the question of what a consumer actually owns when they buy something digital will next be argued.

The point


The mixtape was not a legal grey area that technology happened to close. It was a settled European compromise, and a deliberate one: culture could move between people without permission and without a record, and creators were paid for it through a mechanism that required watching nobody. Europe had built the thing the digital debate is still said to be searching for — a way to pay creators that does not depend on controlling what individuals do with what they own.

Then it adopted, on top of that mechanism, a legal architecture designed elsewhere for a country that had never built one. Every measure since has chipped away at one half of the European compromise while leaving the other half unimproved, and the half that erodes is the one that pays artists. Pirates want copyright reform not because copying should be free of consequence, but because a law written to reward authors has been rebuilt, piece by piece, into a system for deciding who may hand what to whom — and that is a different thing entirely.

Read more on our position, to support the review of the 2019 Copyright in the Digital Single Market Directive in the light of the even more modern advances in technology, below.

Consultation Feedback on CDSM – European PiratesDownload


reshared this

The Pirate Post ha ricondiviso questo.

Why Pirates want Copyright Reform


Someone once spent an evening choosing the order of twelve songs, wrote the titles on the paper insert by hand, and gave the cassette to a person they were hoping to impress. It was a copy. It was made without asking anyone. Across much of Europe it also fell squarely within an exception that lawmakers had gone out of their way to create. That was not an oversight. Germany wrote a private copying exception into its Copyright Act in 1965 and paid for it with a levy on recording equipment and […]
The media in this post is not displayed to visitors. To view it, please go to the original post.

Someone once spent an evening choosing the order of twelve songs, wrote the titles on the paper insert by hand, and gave the cassette to a person they were hoping to impress. It was a copy. It was made without asking anyone. Across much of Europe it also fell squarely within an exception that lawmakers had gone out of their way to create.

That was not an oversight. Germany wrote a private copying exception into its Copyright Act in 1965 and paid for it with a levy on recording equipment and blank media. Most of continental Europe followed the same model, and the arrangement was eventually carried into EU law as the private copying exception in Article 5(2)(b) of the InfoSoc Directive — the 2001 law that still governs copyright across the European Union. Rightsholders were compensated. Nobody had to be watched.

The record industry campaigned against it anyway. “Home taping is killing music” was the British Phonographic Industry’s slogan from 1981, printed on record sleeves under a skull-and-crossbones. It is worth noticing what that campaign did with language: an act that European lawmakers had deliberately permitted, and for which artists were already being paid through the levy, was described as killing. The framing arrived two decades before file-sharing became a mass phenomenon — the first clue that the argument was never really about the copying.

Two answers, and the one that travelled


Europe and the United States faced the same question in the cassette era and answered it in opposite ways.

Europe legalised the copying and paid for it collectively. Levies on recording equipment and blank media flow through collecting societies to authors, performers, publishers and visual artists. According to the global study published by the collecting societies themselves, the system now operates in 31 European countries against four in the whole of the Americas, generates over €1 billion a year worldwide, and costs each consumer between €1.14 and €3.19 annually. In more than twenty countries part of the money funds social and cultural programmes — training, pensions, support for emerging artists — which is why this is better understood as social infrastructure than as a tax. The Court of Justice has held that up to half of such compensation may be directed to those institutions. And the study describes the defining feature of the whole arrangement in a single phrase: the money reaches creators “without monitoring individual users.”

The system has real critics. Europe’s technology industry argues that levies are an outdated instrument that charges buyers for copies many of them never make, applied through tariffs that vary widely and unpredictably between Member States. That is a fair objection to how the levy is calculated. It is not an argument against the principle underneath it.

The United States built nothing comparable. Its Audio Home Recording Act of 1992 did create a royalty, but only on dedicated digital audio recording devices — and when the recording industry tried to apply it to the first MP3 players, the Ninth Circuit held in RIAA v. Diamond Multimedia (1999) that the Rio was not such a device, because it merely made portable what was already on a computer’s hard drive. General-purpose computers fell outside the scheme, and the scheme never recovered. One comparison put annual private copying payments at $1.2 billion in the EU against roughly $1 million in the United States.

With no mechanism to compensate for copying, there was nothing to offer in exchange for permitting it. The American answer to home copying could only be to prevent it. That answer was written into the WIPO Copyright Treaty in 1996, into the Digital Millennium Copyright Act in 1998, and then into European law in the InfoSoc Directive of 2001 — the very directive that codified the private copying exception. Europe kept its levy and imported the lock in the same instrument.

Europeans consequently pay twice. The levy is charged on the device on the assumption that private copies will be made. The DRM on the file then prevents the copy the levy has already paid for.

Canada shows where that combination ends. It adopted the European model in 1997, with a private copying exception for music funded by a levy — but the statute named media rather than devices. When the Copyright Board extended the tariff to the memory inside MP3 players, the Federal Court of Appeal struck it down for want of authority, and the Supreme Court declined to intervene. The levy stayed on blank CDs while the copying moved to phones, and receipts fell from C$38 million in 2004 to C$1.1 million in 2019. Canada then implemented the anti-circumvention rules in 2012 regardless, alongside a set of user exceptions more generous than anything in EU law — format-shifting, backup, and a non-commercial remix exception Europe still lacks — all of which the digital locks override. The lesson is worth stating precisely: the Canadian levy did not fail because collective compensation does not work. It failed because a statute written around cassettes was never rewritten.

The companies that came to operate the distribution layer grew up under the American rule set rather than the European one. Apple’s iTunes Store, Amazon’s Kindle, Google’s YouTube, Netflix and Steam determine the terms on which most Europeans now reach music, books, film and games, with Spotify the notable European exception. The safe harbour that made platforms built on user uploads commercially viable was an American design. Europe’s eventual answer, Article 17, imposes filtering costs — Google says it has invested over $100 million in Content ID alone — that in practice only the largest incumbents can absorb.

Meanwhile the European mechanism erodes. Levy collections across 32 countries have been broadly flat in cash terms since 2017 and have fallen in real terms, because the copying has moved to streaming, where the exception, and the payment attached to it, does not clearly reach.

The copy you cannot give away


Buy a paperback and it is yours. Resell it, lend it, give it away, leave it to someone in a will — copyright does not stand in the way, because the right to control distribution of that copy is exhausted by the first authorised sale. This is not a loophole. Exhaustion is the mechanism that makes libraries, second-hand shops and gifts possible.

Buy the same book as an e-book and none of it holds. In Tom Kabinet (C-263/18, 19 December 2019), the Court of Justice of the European Union ruled that supplying an e-book for permanent download is not “distribution” at all but “communication to the public” — a right that is never exhausted. A second-hand market in e-books is therefore unlawful without the publisher’s permission. The Court had reached the opposite conclusion for downloaded software seven years earlier, in UsedSoft (C-128/11), which tells us the distinction is a policy choice rather than a fact of nature.

No piracy is involved anywhere in this. A reader who paid full price cannot pass the book on. What was removed was not an unauthorised act but an ordinary one.

The lock that outranks the right


The private copying exception still exists on paper. Exercising it is another matter.

The anti-circumvention rules imported in 2001 make it unlawful to defeat a technological protection measure — the family of access controls usually called DRM. Article 6(4) was meant to be the safety valve between the two halves of the directive: Member States should ensure that people can still benefit from certain exceptions even when a work is locked. In practice it applies to a closed list of exceptions, leaves the remedy to national procedures that most Member States never built, and offers nothing at all where a work is streamed on demand.

The result is a right that cannot be exercised. The lock wins, and the lock does not distinguish between an infringing copy and a lawful one — it cannot, because it does not know what the user intends.

Portugal is the exception that proves the point. Law 36/2017 permits circumvention where the purpose is to exercise a lawful exception, and forbids DRM on public domain works and on works funded by public money. Campaigners there needed fifteen years to win a right that the Directive had always claimed to guarantee.

Enforcement moved into the pipes


Article 17 of the 2019 Copyright in the Digital Single Market Directive made large platforms liable for what their users upload unless they make best efforts to prevent it. It was argued for as a fix to the “value gap” between what platforms earn and what rightsholders receive. What it requires in practice is automated inspection of everything anyone posts.

The Court of Justice upheld the provision in Poland v Parliament and Council (C-401/19, 26 April 2022), but not in the form several governments wanted. It rejected the argument that content could be blocked systematically at upload so long as users could appeal afterwards, and held that safeguards must operate before blocking, not after. A filter that cannot tell a quotation from an infringement may not be deployed as though it can.

That judgment is a genuine win for user rights. It is also an admission about the mechanism: the default behaviour of automated enforcement is to remove lawful speech, and it took the EU’s highest court to say so.

When there is no copy at all


The clearest case is the one currently on the table.

The European citizens’ initiative “Stop Destroying Videogames” was submitted to the Commission on 26 January 2026with 1,294,188 verified signatures. It asked for something modest: that a game sold to the public should be left in a playable state when the publisher stops supporting it. On 16 June 2026 the Commission replied that it cannot at this stage propose such an obligation, and offered instead to convene the industry on a voluntary code of conduct by the end of 2026.

The background to that request is Ubisoft’s racing game The Crew, delisted in December 2023 and switched off three months later. Buyers then found their licences revoked outright. In the litigation that followed — including a case brought by the French consumer association UFC-Que Choisir — Ubisoft’s defence was that customers had never bought a copy at all, only a limited licence to access a service.

That defence is legally coherent, and that is precisely the problem. When the transaction is a licence rather than a copy, there is nothing to lend, resell, archive or hand to anyone. The unmonitored exchange of culture between people does not have to be prohibited. It simply has nothing left to move.

The case on the other side


It deserves stating properly. Creators need income, and an unlimited right to copy and redistribute commercially would remove the market that pays them. Rightsholders cannot realistically police millions of individual users, so building enforcement into infrastructure is a rational response to a real problem. Licensed streaming has been a genuine success on its own terms: cheap, legal and convenient, and its defenders can fairly point out that it did more to draw people away from unauthorised downloading than a decade of lawsuits ever managed.

None of that is in dispute. The question is what the mechanisms actually deliver.

Extending the term of protection for sound recordings from 50 to 70 years, as the EU did in 2011, prevents no unauthorised copying whatsoever — it can only extend control over recordings that already exist. A study prepared for the European Parliament sets out the recurring criticism: the benefit falls mainly to labels and a small number of successful performers, while ordinary session musicians have little bargaining power over the contracts that determine what reaches them. The levy that paid for the mixtape paid artists directly. Digital rights management, upload filters and licence-only distribution do not pay them more; they change who decides on what terms a work reaches the public.

That is the reform question. Not whether creators should be paid — they should, and better than they currently are — but who holds the switch.

What reform actually asks for


The European Pirates’ programme states the position directly:

  • Non-commercial use is protected, not merely tolerated. “Copying, storing, using, and providing access to literary and artistic works for non-commercial purposes must not only be legalized but protected by law.” Non-commercial file sharing should be allowed.
  • Shorter commercial terms. The term of the commercial monopoly should be further shortened, exclusive rights must be limited in time and scope, and neither may be expanded retrospectively.
  • Derivative and everyday uses carved out. Remixing, parody, quotation and sampling exempt from the commercial monopoly; derivative works permitted by default, with exceptions written explicitly into law; linking never an infringement.
  • Exceptions made mandatory across the EU, so that a lawful use in one Member State is not an infringement in the next.
  • Public money, public access. Cultural heritage digitised and made available free of charge; an end to the privatisation of profits from publicly funded works.
  • Transparency from collecting societies, and fair contract terms for the artists on whose behalf all of this is said to be done.

Four further steps follow from the mechanisms described above, and are where the argument now runs. A right to circumvent DRM where the purpose is a lawful one, together with a ban on DRM applied to public domain works and works funded from public money, on the Portuguese model. Digital exhaustion, so that a purchased digital copy can be resold, lent, given away and inherited like its physical equivalent. An end-of-life duty, so that a work withdrawn from sale is left usable rather than switched off — the ask that 1.29 million Europeans have already signed. And a collective compensation mechanism that survives the shift away from physical media, because the European alternative to controlling users was always to pay creators without watching them, and that half of the bargain is the half currently being allowed to lapse.

Some of this is live now. The Commission’s Digital Fairness Act, expected in the second half of 2026, is where the question of what a consumer actually owns when they buy something digital will next be argued.

The point


The mixtape was not a legal grey area that technology happened to close. It was a settled European compromise, and a deliberate one: culture could move between people without permission and without a record, and creators were paid for it through a mechanism that required watching nobody. Europe had built the thing the digital debate is still said to be searching for — a way to pay creators that does not depend on controlling what individuals do with what they own.

Then it adopted, on top of that mechanism, a legal architecture designed elsewhere for a country that had never built one. Every measure since has chipped away at one half of the European compromise while leaving the other half unimproved, and the half that erodes is the one that pays artists. Pirates want copyright reform not because copying should be free of consequence, but because a law written to reward authors has been rebuilt, piece by piece, into a system for deciding who may hand what to whom — and that is a different thing entirely.

Read more on our position, to support the review of the 2019 Copyright in the Digital Single Market Directive in the light of the even more modern advances in technology, below.

Consultation Feedback on CDSM – European PiratesDownload

reshared this

The Pirate Post ha ricondiviso questo.

Würde die AfD ein Bundesland regieren, hätte sie eine Reihe von Möglichkeiten, den Sicherheitsapparat in ihrem Sinne neu auszurichten. Doch Demokrat*innen stehen dem nicht hilflos gegenüber. Wir haben mit dem Rechtswissenschaftler Markus Thiel darüber gesprochen, welche Gefahren eine potenzielle Machtübernahme birgt und was man dagegen tun kann. netzpolitik.org/2026/polizei-u…
in reply to netzpolitik.org

Allein das es notwendig erscheint, sich solche Fragen zu stellen, macht deutlich, dass in diesem Land etwas gewaltig schiefläuft. Etwas das jahrelang, wenn nicht jahrzehntelang, von praktisch allen Stellen unterschätzt oder ignoriert wurde. Offensichtlich waren die demokratische Staatsform, ein Grundgesetz und ein öffentlicher Rundfunk nicht genug, um das Ziel zu erreichen, nie wieder Faschismus zuzulassen. Wir müssen wohl wieder lernen, für Demokratie und Frieden zu kämpfen.
The Pirate Post ha ricondiviso questo.

☕ CYBERBRIEFING MATTUTINO — Martedì 18 agosto 2026

👉 Leggi tutti gli aggiornamenti delle ultime 24 ore:
ilpuntocyber.rfeed.it/article.…

#newsletter #cybersecurity
@informatica

Bastian’s Night #490 August, 20th


Every Thursday of the week, Bastian’s Night is broadcast from 21:30 CEST/DST.

Bastian’s Night is a live talk show in German with lots of music, a weekly round-up of news from around the world, and a glimpse into the host’s crazy week in the pirate movement.


If you want to read more about @BastianBB: –> This way


piratesonair.net/bastians-nigh…

Elezioni e Politica 2026 reshared this.

ICYMI: Updates from the 8/16 Meeting


ICYMI

Arizona – The latest AZPP meeting covered several key items: Jane Gardner stepped down as YPUSA representative due to workload, with the group thanking her for her service. The team spent most of the meeting reviewing and updating the bylaws (Version 3 now in the resources chat), and discussed fundraising needs for the write-in campaign and cutlass maps. They also talked about a possible 2027 PNC conference bid and building a Tucson itinerary for out-of-town members.

On the campaign side, yard signs need to be finalized soon (with Gloo Factory in Tucson as a leading option for premade signs), and Blase Henry’s No Kings Day Speech video ad is now live. Game Night got the green light after Twitch approval, with a possible first stream of AC4: Black Flag on Tuesday. The Phoenix event idea was tabled for next time, and outreach/committee work continues on PayloadCMS.

Financial – A financial policy and procedure was introduced by Ohio and unilaterally adopted. The document, among other things, commits the Treasurer to report the financial standings of the USPP to the board at least once a month.

Illinois – The Midlothian Village Board of Trustees campaign kicked off with an announcement on Friday, officially launching the first campaign since the 2022 attempt. This will also be the first race the ILPP have an invested interest in since the 2023 Chicago Mayoral Race, where the Chicagoland Pirate Party endorsed a slate of four candidates before the runoff. That announcement can be found here.

Pirate National Committee – The Pirate National Committee voted to cancel select upcoming meetings, all related to potential conflict with holiday and travel, through to the Pirate National Conference. That list of cancelled dates are as follows:

  • Sept. 6th, 2026 – Labor Day Weekend
  • Nov. 29th, 2026 – Thanksgiving Weekend
  • Dec. 27th, 2026 – Christmas Weekend
  • Jan. 3rd, 2027 – NYE Weekend
  • Feb. 14th, 2027 – SuperBowl + St. Valentine’s Day
  • March 28th, 2027 – Easter Sunday
  • May 9th, 2027 – Mother’s Day
  • May 30th, 2027 – Memorial Day Weekend

Historically, the Pirate National Conference has taken place the first weekend of June, meaning there will likely not be a meeting June 6th, 2027 (hey, that’s the U.S. Pirate Party’s 21st birthday!), but that is currently not an officially voted on cancellation.

Further cancellations regarding holidays will be discussed after a new board is elected during the 2027 Pirate National Conference.

Pirate National Conference – the 2027 Pirate National Conference location is currently being discussed, with an Etherpad created to answer any questions and hear all suggests Pirates and Pirate supporters might have for the conference. The favorites are currently Baltimore, Mobile and Tucson.

Vetting Committee – A Vetting Committee has been formed to properly vet any candidates seeking the endorsement of the Pirate National Committee and the United States Pirate Party. The Committee will be comprised of the Captain, Vice Captain, PR Director, a Captain Emeritus and two reps selected from PNC.


That’s all from last night! Or is it? Catch last night’s meeting here to find out.


uspirates.org/icymi-updates-fr…

Elezioni e Politica 2026 reshared this.

Notice on Future Cancellations of Select PNC Meetings


August 17

During the August 16th Pirate National Committee meeting, it was voted on that the PNC would not meet on the following days:

  • Sept. 6th, 2026 – Labor Day Weekend
  • Nov. 29th, 2026 – Thanksgiving Weekend
  • Dec. 27th, 2026 – Christmas Weekend
  • Jan. 3rd, 2027 – NYE Weekend
  • Feb. 14th, 2027 – SuperBowl + St. Valentine’s Day
  • March 28th, 2027 – Easter Sunday
  • May 9th, 2027 – Mother’s Day
  • May 30th, 2027 – Memorial Day Weekend

Historically, the Pirate National Conference has taken place the first weekend of June, meaning there will likely not be a meeting June 6th, 2027 (hey, that’s the U.S. Pirate Party’s 21st birthday!), but that is currently not an officially voted on cancellation.

Further cancellations in regards to holidays will be discussed after a new board is elected during the 2027 Pirate National Conference.


uspirates.org/notice-on-future…

Elezioni e Politica 2026 reshared this.

The Pirate Post ha ricondiviso questo.

Ah, der überall als harmlos und Vorzeige-CDUler gefeierte Herr Günther will also das schärfste Polizeigesetz, das Deutschland zu bieten hat. Wie überraschend, nicht.


„Direkt aus dem Roman 1984“: Das neue Polizeigesetz von Schleswig-Holstein soll eines der härtesten Deutschlands werden. Die Gruppe @freiheitsfoo analysiert es in einem 22seitigen Gutachten. Ihr Ergebnis: Der Entwurf ist eine Bedrohung für die Demokratie. netzpolitik.org/2026/polizeige…

Questa voce è stata modificata (4 settimane fa)
The Pirate Post ha ricondiviso questo.

The media in this post is not displayed to visitors. To view it, please go to the original post.

The media in this post is not displayed to visitors. To view it, please go to the original post.

✨ CVSS 10 senza PoC pubblico: la corsa contro il tempo sulla falla critica di SAP Commerce Cloud
#CyberSecurity
insicurezzadigitale.com/cvss-1…

@informatica


CVSS 10 senza PoC pubblico: la corsa contro il tempo sulla falla critica di SAP Commerce Cloud


Si parla di:
Toggle

SAP ha rilasciato la patch l’11 agosto. Tre giorni dopo, gli honeypot dell’azienda di threat intelligence Defused hanno registrato i primi tentativi di sfruttamento. Nessun proof-of-concept pubblico, nessun writeup tecnico in circolazione: qualcuno ha fatto reverse engineering della patch – o conosceva già la falla – più in fretta di quanto la community difensiva riuscisse a reagire. La vulnerabilità, CVE-2026-58231, ha un CVSS di 10.0: il massimo possibile, per una piattaforma usata da alcuni dei più grandi retailer al mondo.

La falla: un client di autenticazione dimenticato


CVE-2026-58231 risiede nell’estensione Data Hub Adapter di SAP Commerce Cloud, la piattaforma e-commerce nota fino a pochi anni fa come SAP Hybris. Secondo l’advisory ufficiale, controlli di autorizzazione insufficienti e una validazione dell’input carente permettono a un attaccante non autenticato di “abusare di un client di autenticazione predefinito” e sottoporre input appositamente costruiti al sistema. Il risultato è esecuzione di codice arbitraria in remoto (RCE), raggiungibile con complessità di attacco bassa e senza alcuna credenziale: la combinazione che giustifica il punteggio massimo di gravità.

SAP Commerce Cloud non è un prodotto di nicchia: è l’infrastruttura e-commerce dietro cataloghi, checkout e gestione ordini di catene retail, produttori e distributori a livello globale. Una RCE non autenticata su questo layer significa accesso diretto a dati di pagamento, anagrafiche clienti e, potenzialmente, un punto di ingresso verso i sistemi ERP interni a cui Commerce Cloud è tipicamente integrato.

Tre giorni: dalla patch al primo probe


La finestra temporale è ciò che rende questo caso degno di nota. SAP ha pubblicato la correzione l’11 agosto 2026 tramite la security note 3771065. Il 14 agosto, Defused ha osservato sui propri sensori i primi tentativi di exploitation, comunicandolo pubblicamente su X. Shadowserver Foundation, che traccia costantemente l’esposizione internet di SAP Commerce Cloud tramite fingerprinting, segnala oltre 4.200 indirizzi IP con firma compatibile con la piattaforma – la superficie di attacco potenziale su cui gli scanner automatizzati si stanno già muovendo.

Tre giorni sono un tempo sufficiente per un diffing accurato della patch (confrontando i binari o i sorgenti pre e post correzione per isolare il codice modificato e dedurre la vulnerabilità), ma insolitamente rapido per la complessità tipica delle applicazioni enterprise Java di SAP. Onapsis, società specializzata nella sicurezza degli ambienti SAP, non esclude che gli attaccanti stiano semplicemente testando in massa endpoint esposti sperando di trovare installazioni non ancora patchate, un pattern comune subito dopo il rilascio di patch critiche per software enterprise molto diffuso.

Il precedente che preoccupa i difensori


SAP Commerce Cloud e le altre piattaforme della galassia SAP non sono nuove a questo tipo di attenzione. Nel maggio 2025, la vulnerabilità gemella CVE-2025-31324 in SAP NetWeaver Visual Composer fu sfruttata da almeno un cluster di attività riconducibile a operazioni di cyberspionaggio cinesi per compromettere oltre 580 sistemi critici a livello globale, secondo l’analisi di EclecticIQ, prima che gruppi ransomware si aggiungessero alla caccia agli endpoint non patchati. Lo schema si ripete con regolarità: le vulnerabilità critiche nei sistemi SAP attirano prima attori sofisticati orientati all’intelligence, interessati all’accesso silenzioso a dati finanziari e di supply chain, e solo in un secondo momento la criminalità informatica opportunistica in cerca di un punto d’appoggio da rivendere o da usare per il deployment di ransomware.

Per ora non ci sono attribuzioni pubbliche legate ai probe osservati su CVE-2026-58231: i dati disponibili derivano da honeypot e da telemetria di scansione, non da un’intrusione confermata con post-exploitation osservata. Ma la storia recente di SAP suggerisce che la finestra tra probe automatizzati e sfruttamento mirato da parte di attori più sofisticati tende a essere breve.

Cosa devono fare i team di sicurezza


  • Applicare immediatamente la security note SAP 3771065, aggiornando ai livelli di patch corretti per Commerce Cloud
  • Dove l’aggiornamento immediato non è possibile, ricompilare e ridistribuire la versione corretta appena disponibile una finestra di manutenzione
  • Come mitigazione temporanea, configurare un IP Filter Set per limitare l’accesso all’endpoint del Data Hub Adapter alle sole reti fidate
  • Verificare nei log applicativi richieste anomale dirette agli endpoint di Data Hub Adapter nei giorni successivi all’11 agosto
  • Monitorare gli avvisi di Shadowserver e della propria threat intelligence per indicatori di scansione mirata verso il proprio range IP
  • Trattare qualsiasi istanza Commerce Cloud esposta a Internet e non patchata come potenzialmente già compromessa, avviando attività di hunting sui sistemi integrati (ERP, data warehouse, gateway di pagamento)

Il messaggio per chi gestisce ambienti SAP è lo stesso che si ripete a ogni ciclo di patch critiche in software enterprise molto esposto: il tempo tra il rilascio di una correzione e la sua trasformazione in arma da parte degli attaccanti si sta comprimendo, e in questo caso specifico non c’è stato nemmeno bisogno di un proof-of-concept pubblico per accorciarlo ulteriormente.

Dettagli tecnici

# Vulnerabilità
CVE-2026-58231
CVSS 3.1: 10.0 (Critico)
Componente: SAP Commerce Cloud - estensione Data Hub Adapter
Tipo: Autorizzazione impropria + input validation carente -> RCE non autenticata

# Timeline
2026-08-11  Rilascio patch (SAP Security Note 3771065)
2026-08-14  Primi tentativi di exploitation osservati (Defused, honeypot)
2026-08-17  Oltre 4.200 IP con fingerprint SAP Commerce Cloud tracciati da Shadowserver

# Mitigazione
- Applicare Security Note 3771065
- IP Filter Set sull'endpoint Data Hub Adapter come workaround temporaneo

# Precedente correlato
CVE-2025-31324 - SAP NetWeaver Visual Composer, sfruttata da cluster
di cyberspionaggio cinesi su 580+ sistemi critici (maggio 2025)

Fonti: The Hacker News, BleepingComputer, SecurityAffairs, GBHackers, Onapsis, Shadowserver Foundation, EclecticIQ.

The Pirate Post ha ricondiviso questo.

Ein Künstler zeigt mit einem besonderen Hemd, wie automatische Videoüberwachung ausgetrickst werden kann. Es geht ihm dabei weniger um den Trick an sich, als um grundsätzliche Kritik an der invasiven und intransparenten Technologie.

netzpolitik.org/2026/protest-k…

The Pirate Post ha ricondiviso questo.

Juhu, ich bin nicht der Einzige, der das neue Polizeigesetz von Schleswig-Holstein für superdystopisch hält! @freiheitsfoo hat eine Stellungnahme dazu veröffentlicht, die angenehm klare Worte findet. "Direkt aus dem Roman 1984" ist da nur eines von sehr vielen sehr starken Zitaten. Wer nicht gleich die ganzen 22 Seiten lesen mag, kann sich hier meine Zusammenfassung geben: netzpolitik.org/2026/polizeige…
The Pirate Post ha ricondiviso questo.

„Direkt aus dem Roman 1984“: Das neue Polizeigesetz von Schleswig-Holstein soll eines der härtesten Deutschlands werden. Die Gruppe @freiheitsfoo analysiert es in einem 22seitigen Gutachten. Ihr Ergebnis: Der Entwurf ist eine Bedrohung für die Demokratie. netzpolitik.org/2026/polizeige…
in reply to netzpolitik.org

In Schleswig-Holstein regieren CDU und Grüne. Sie beide setzen diese rechtsextreme Politik um. Nur der CDU du Schuld zu geben, wie einige hier in den Kommentaren, ist Realitätsverleugnung.

Da spielt sicherlich auch mit rein, dass viele Grünen-Wähler/innen sich nicht eingestehen wollen, dass sie selbst so auf Wahlkampfversprechen und Selbstinszenierung Partei hereinfallen wie CDU-Wähler/innen.

The Pirate Post ha ricondiviso questo.

The media in this post is not displayed to visitors. To view it, please go to the original post.

CVE-2026-55040: bypass di autenticazione in SharePoint sotto attacco attivo, ecco come proteggersi
#tech
spcnet.it/cve-2026-55040-bypas…
@informatica


CVE-2026-55040: bypass di autenticazione in SharePoint sotto attacco attivo, ecco come proteggersi


Quando una vulnerabilità critica in un prodotto enterprise passa dalla ricerca accademica allo sfruttamento attivo nel giro di poche settimane, il tempo a disposizione dei team IT per applicare la patch si riduce drasticamente. È esattamente quello che sta succedendo con CVE-2026-55040, un bypass di autenticazione in Microsoft SharePoint Server che, dopo la pubblicazione di un proof-of-concept dettagliato, è ora oggetto di tentativi di exploitation attiva da IP distribuiti in più paesi. Per chi gestisce ambienti SharePoint on-premises, capire il meccanismo dell’attacco è il primo passo per dare priorità corretta alla remediation.

Cos’è CVE-2026-55040


La vulnerabilità, con punteggio CVSS 3.1 pari a 9.1 (critico) e classificata come CWE-1390 (Weak Authentication), risiede nella pipeline di validazione dei token JWT usata da SharePoint per l’autenticazione server-to-server (S2S). Il risultato pratico è che un attaccante non autenticato, conoscendo l’identificativo di un utente (il SID di Active Directory o lo UPN), può forgiare un token che SharePoint accetta come legittimo — arrivando fino all’impersonificazione di un account amministrativo.

Sono colpite tre versioni principali del prodotto:

  • SharePoint Server Subscription Edition (build 16.0.19725.20434 e precedenti)
  • SharePoint Server 2019 (build 16.0.10417.20175 e precedenti)
  • SharePoint Enterprise Server 2016 (build 16.0.5561.1001 e precedenti)


Come funziona il bypass


L’analisi tecnica pubblicata da Rapid7 Labs descrive una catena di quattro debolezze concatenate nel processo di verifica dei token Bearer S2S, che nel loro insieme permettono di costruire un token accettato senza una firma crittografica valida: un header che dichiara un algoritmo di firma “none”, l’uso del certificato STS di SharePoint stesso come chiave di firma anziché come verificatore, un certificato non correttamente validato contro TrustedSecurityTokenServices, e — punto più critico — la firma del token che di fatto non viene mai verificata end-to-end.

In pratica, un attaccante che conosce (o enumera) l’identificativo di un utente può costruire un token che SharePoint tratta come proveniente da un servizio fidato, ottenendo l’accesso alle API e alle risorse del sito con i privilegi di quell’utente — amministratore incluso, se il SID target è quello giusto.

Perché è particolarmente pericolosa


Tre fattori la rendono un rischio prioritario rispetto alla media delle CVE mensili:

  • Nessuna autenticazione richiesta: l’attaccante non ha bisogno di credenziali valide, solo della conoscenza (o della capacità di enumerare) l’identificativo dell’utente target.
  • PoC pubblico e riproducibile: la disponibilità di un exploit funzionante abbassa drasticamente la barriera tecnica per chi vuole sfruttarla, come dimostra l’aumento dei tentativi osservati nei giorni successivi alla pubblicazione.
  • Superficie di attacco diretta: qualsiasi istanza SharePoint on-premises esposta a Internet, anche solo per l’accesso remoto di dipendenti o partner, è un bersaglio potenziale.


Cosa mostra la telemetria degli attacchi


Secondo i dati di tracking condivisi dalla community di threat intelligence (KEVIntel), i primi tentativi di sfruttamento risalgono al 19 luglio 2026, con una intensificazione marcata nei giorni successivi alla pubblicazione del PoC — otto tentativi registrati solo tra il 12 e il 13 agosto, provenienti da otto indirizzi IP distinti localizzati tra Hong Kong, Giappone, Paesi Bassi, Taiwan e Stati Uniti. Un pattern tipico delle campagne opportunistiche che seguono la pubblicazione di un exploit pubblico: scansioni automatizzate su larga scala alla ricerca di istanze non ancora aggiornate.

Patch e mitigazioni


Microsoft ha già rilasciato gli aggiornamenti correttivi; il primo passo per qualsiasi team che gestisce SharePoint on-premises è verificare di aver applicato i seguenti KB in base alla versione in uso:

SharePoint Server Subscription Edition → KB5002882
SharePoint Server 2019                 → KB5002883
SharePoint Enterprise Server 2016      → KB5002891

Oltre all’applicazione della patch, che resta la mitigazione principale, vale la pena eseguire una checklist di hardening più ampia:
  • Verificare la build corrente tramite l’interfaccia di amministrazione centrale o PowerShell (Get-SPFarm | Select BuildVersion), confrontandola con le versioni corrette rilasciate da Microsoft.
  • Analizzare i log IIS e ULS alla ricerca di pattern di richieste anomale verso gli endpoint di autenticazione S2S, in particolare token Bearer con struttura o claim inusuali.
  • Ridurre l’esposizione diretta a Internet dei server SharePoint on-premises dove non strettamente necessario, preferendo l’accesso tramite VPN o reverse proxy con autenticazione aggiuntiva.
  • Segmentare la rete in modo che un’eventuale compromissione del front-end SharePoint non dia accesso diretto ad altri sistemi interni.
  • Rivedere gli account con privilegi elevati su SharePoint, applicando il principio del privilegio minimo e monitorando le attività degli account amministrativi per individuare comportamenti anomali successivi alla finestra di esposizione.


Conclusione


CVE-2026-55040 è un promemoria diretto di quanto velocemente un bypass di autenticazione ben documentato possa trasformarsi in sfruttamento attivo su scala. Per chi amministra ambienti SharePoint on-premises, la priorità immediata è verificare lo stato delle patch KB5002882/KB5002883/KB5002891 e, in assenza di conferma dell’aggiornamento, trattare l’istanza come potenzialmente compromessa fino a prova contraria — controllando log di autenticazione e attività degli account amministrativi nella finestra temporale in cui la vulnerabilità è rimasta senza patch.

Fonti: Petri IT Knowledgebase, Rapid7 Labs, The Hacker News


The Pirate Post ha ricondiviso questo.

The media in this post is not displayed to visitors. To view it, please go to the original post.

PHP-FPM in produzione: perché pm static batte dynamic e come calcolare pm.max_children
#tech
spcnet.it/php-fpm-in-produzion…
@informatica


PHP-FPM in produzione: perché pm static batte dynamic e come calcolare pm.max_children


Chi gestisce uno stack LEMP o LAMP in produzione conosce bene il dilemma: PHP-FPM va lasciato “respirare” con un process manager dinamico, oppure conviene bloccare tutto su un numero fisso di worker? La risposta non è scontata, e sbagliarla ha un costo diretto in latenza durante i picchi di traffico. Vediamo perché, per molti carichi di lavoro, pm static è la scelta più solida — e come calcolare i parametri giusti senza affidarsi a numeri presi a caso da qualche post su Stack Overflow.

I tre process manager di PHP-FPM


PHP-FPM gestisce i processi worker attraverso la direttiva pm nel file di pool (tipicamente /etc/php/8.x/fpm/pool.d/www.conf), che può assumere tre valori:

  • dynamic: il numero di processi figlio oscilla tra pm.min_spare_servers e pm.max_spare_servers, fino al tetto di pm.max_children. È il default su molte distribuzioni ed è pensato per bilanciare consumo di RAM e capacità di risposta.
  • ondemand: i processi vengono creati solo quando arriva una richiesta e vengono terminati dopo pm.process_idle_timeout secondi di inattività. Nessun worker “a riposo” occupa memoria, ma ogni nuovo processo comporta un fork che ha un costo in latenza.
  • static: il numero di worker è fissato esattamente a pm.max_children, tutti avviati all’avvio del servizio e mai terminati. Nessun fork a runtime, nessuna sorpresa.

La differenza pratica tra dynamic/ondemand e static si vede sotto carico: ogni volta che PHP-FPM deve forkare un nuovo processo per rispondere a un picco di richieste, quel fork costa tempo di CPU e, sotto pressione, può accodare le richieste in ingresso nel socket backlog. Su un server che serve traffico costante e prevedibile, questo overhead è puro spreco.

Perché “dynamic” può ingannare


Il process manager dinamico sembra la scelta “intelligente” perché si adatta al carico. In realtà introduce una variabile in più proprio nei momenti in cui non la vorresti: durante un picco di traffico, quando i worker spare non bastano più, PHP-FPM deve forkarne di nuovi mentre il sistema è già sotto stress. Il risultato è un aumento di latenza percepibile negli access log, spesso confuso con un problema del database o del codice applicativo quando in realtà è il process manager stesso a introdurre il collo di bottiglia.

Quando usare pm static


La regola pratica è semplice: se il server ha traffico sostenuto e memoria sufficiente per tenere tutti i worker sempre attivi, static è la scelta giusta. Se invece si gestiscono più pool PHP-FPM su un host condiviso con memoria limitata (tipico di ambienti multi-tenant o hosting condiviso), ondemand resta preferibile perché libera RAM quando il traffico cala.

; /etc/php/8.3/fpm/pool.d/www.conf
pm = static
pm.max_children = 40
pm.max_requests = 1000

Da notare pm.max_requests: anche in modalità static conviene non lasciarlo a 0 (illimitato). Un valore alto — 1000 richieste è un buon punto di partenza — fa sì che ogni worker venga riciclato periodicamente, mitigando eventuali memory leak nelle librerie PHP senza introdurre l’overhead di respawn continui tipico della modalità dinamica.

Come calcolare pm.max_children senza indovinare


Il valore corretto di pm.max_children dipende da quanta RAM è disponibile e da quanta ne consuma mediamente un singolo worker PHP-FPM — un dato che varia moltissimo in base al framework (un’app Symfony con Doctrine pesa parecchio di più di uno script WordPress minimale).

Per misurare il consumo medio reale dei worker già in esecuzione:

ps --no-headers -o rss -C php-fpm8.3 | awk '{ sum += $1; n++ } END { print sum/n/1024 " MB" }'

Questo comando interroga la RSS (Resident Set Size) di tutti i processi php-fpm attivi e ne calcola la media in MB. Con quel numero in mano, la formula diventa:
pm.max_children = (RAM dedicata a PHP-FPM in MB) / (RSS medio per worker in MB)

Ad esempio, su un server con 6 GB di RAM riservati al pool PHP-FPM e worker che consumano in media 60 MB ciascuno:
6000 MB / 60 MB ≈ 100 worker

È fondamentale non allocare a PHP-FPM tutta la RAM disponibile sulla macchina: bisogna lasciare margine per il sistema operativo, il web server (Nginx o Apache in front-end), la cache degli oggetti (Redis/Memcached) e, se presente sullo stesso host, il database. Una regola prudente è dedicare a PHP-FPM non più del 60-70% della RAM totale su un server dedicato al solo stack web.

Monitoraggio continuo, non solo calcolo una tantum


Il consumo medio per worker cambia nel tempo — un deploy che introduce una libreria pesante, una query N+1 che gonfia il footprint di memoria, o semplicemente un aumento del traffico su endpoint più complessi. Vale la pena schedulare periodicamente il comando ps sopra (ad esempio via cron con output su un file di log, oppure integrandolo in un dashboard di monitoring come Netdata o Prometheus con node_exporter) per verificare che il valore di pm.max_children resti coerente con la realtà, invece di scoprirlo durante un incidente in produzione.

Un altro segnale da tenere d’occhio è il log di PHP-FPM stesso: se compaiono righe del tipo

WARNING: [pool www] server reached pm.max_children setting (40), consider raising it

significa che il tetto è stato raggiunto e le richieste in eccesso stanno finendo in coda sul socket, con conseguente aumento del tempo di risposta. È il segnale più diretto per capire se il dimensionamento fatto a tavolino regge davvero sotto carico reale.

Conclusione


Non esiste un process manager “giusto” in assoluto: la scelta dipende dal profilo di traffico e dalla disponibilità di memoria. Ma per la maggior parte dei server di produzione con traffico prevedibile e risorse dedicate, pm static abbinato a un pm.max_children calcolato sul consumo reale di RSS — e non copiato da un tutorial generico — elimina una fonte di latenza spesso invisibile finché non arriva il picco di traffico che la rende evidente. Vale la pena rivedere la configurazione dei propri pool PHP-FPM con questo approccio, misurando prima di decidere.

Fonte: LinuxBlog.io — PHP-FPM tuning: Using ‘pm static’ for max performance


The Pirate Post ha ricondiviso questo.

The media in this post is not displayed to visitors. To view it, please go to the original post.

Microsoft Is Merging Consumer and Enterprise Copilot — Security Teams Should Watch the Seams
#CyberSecurity
securebulletin.com/microsoft-i…
The Pirate Post ha ricondiviso questo.

The media in this post is not displayed to visitors. To view it, please go to the original post.

AWS Sets a Multi-Year Countdown to Kill Off Email-Based Certificate Validation
#CyberSecurity
securebulletin.com/aws-sets-a-…
The Pirate Post ha ricondiviso questo.

Werbefirmen und Zeitungsverlagen war Apples Tracking-Schutz für iPhones schon lange ein Dorn im Auge. Nun hat das Kartellamt entschieden, dass der Konzern das System deutlich anpassen muss. Für Nutzer:innen bedeutet das wohl längere und kompliziertere Einwilligungs-Abfragen. Das Wort "Tracking" dürfen diese nicht mehr enthalten - weil es zu abschreckend sein könnte.

netzpolitik.org/2026/bundeskar…

in reply to netzpolitik.org

The media in this post is not displayed to visitors. To view it, please go to the original post.

Ist Euch das auch schon aufgefallen?
Zusätzlich zur paywall kommt auf Websites ein Banner von google...
Ist mir vor ca. 4 Wochen das erste Mal aufgefallen.
Hängt das eventuell auch mit dieser "Trackingfreiheit" zusammen?
Ich nutze absolut NICHTS von Google, habt Ihr da vielleicht eine Idee?
in reply to netzpolitik.org

Für mich wird hier der Teufel mit dem Beelzebub ausgetrieben:

Was Apple als "Dienst an einer exklusiven Kundschaft" vermarktet ist in erster Linie eins: proprietär und intransparent.

Denn über die Loyalität des Konzerns zu seinen Nutzenden kann man religiös streiten - und das Grundproblem übersehen:

In beiden Systemen sind die Nutzer:innen entmündigtes, ohnmächtiges Fußvolk - auch wenn Apple Narrative und Maßnahmen entwickelt hat, damit sich dessen Untertanen besser fühlen.

The Pirate Post ha ricondiviso questo.

The media in this post is not displayed to visitors. To view it, please go to the original post.

Shell Launches Investigation After Cl0p Extortion Group Claims Theft of Nearly 90GB of Internal Data
#CyberSecurity
securebulletin.com/shell-launc…
The Pirate Post ha ricondiviso questo.

The media in this post is not displayed to visitors. To view it, please go to the original post.

Legacy VNC Login on macOS Screen Sharing Could Hand Attackers a Root Shell
#CyberSecurity
securebulletin.com/legacy-vnc-…
The Pirate Post ha ricondiviso questo.

The media in this post is not displayed to visitors. To view it, please go to the original post.

Microsoft Sets Hard Deadline to Kill SMS and Voice Login Codes in Entra ID, Pushes Passkeys Instead
#CyberSecurity
securebulletin.com/microsoft-s…
The Pirate Post ha ricondiviso questo.

The media in this post is not displayed to visitors. To view it, please go to the original post.

Attackers Race to Weaponize Maximum-Severity SAP Commerce Cloud Flaw Within Days of Patch
#CyberSecurity
securebulletin.com/attackers-r…
The Pirate Post ha ricondiviso questo.

The media in this post is not displayed to visitors. To view it, please go to the original post.

Security Leaders Warn the ‘Agentic Attacker’ Has Arrived After AI Models Reportedly Breached Hugging Face on Their Own
#CyberSecurity
securebulletin.com/security-le…