Lorenzo ha ricondiviso questo.

mastodon - Collegamento all'originale
The media in this post is not displayed to visitors. To view it, please go to the original post.

Two design companies, S&A and Playtype, have created a font that fools AI scrappers and protects copyrighted text

s-a.website/projects/shieldfon…

Seneda & Abrucio

A geek and a backpacker who create ideas that make headlines.
s-a.website
Lorenzo ha ricondiviso questo.

mastodon - Collegamento all'originale

Researchers have found a new version of SiliVaccine, North Korea's internal homegrown antivirus engine. Well, not that "homegrown" actually because the engine works on top of ClamAV and Malheur

nkinternet.com/2026/08/01/a-ne…

A New SiliVaccine: North Korea’s Antivirus

In 2018, Check Point published a deep dive into SiliVaccine, North Korea’s homegrown antivirus, and found its scanning engine was built on stolen, decade-old Trend Micro code. Check Point als…
nick (North Korean Internet)

reshared this

Lorenzo ha ricondiviso questo.

mastodon - Collegamento all'originale

A mysterious hacker has stolen more than $70 million worth of Bitcoin from Coldcard hardware wallets: bitcoinmagazine.com/news/coldc…

The attackers exploited a vulnerability in the wallet's firmware that produced weak and easy to guess private keys: blog.coinkite.com/entropy-tech…

Technical Deep Dive into the Entropy Issue

Coinkite Inc. (COINKITE Blog)

reshared this

Lorenzo ha ricondiviso questo.

bridgy-fed - Collegamento all'originale
The media in this post is not displayed to visitors. To view it, please go to the original post.
Anthropic reveals its own rogue-model incidents: Due to human error in configuring a test environment, three Claude models hacked outside organizations during CTF exercises. Two proceeded despite realizing they were on the internet. www.anthropic.com/news/investi...

Investigating three real-world incidents in our cybersecurity evaluations

In a review of our cybersecurity evaluation transcripts, we found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real…
www.anthropic.com

reshared this

Lorenzo ha ricondiviso questo.

mastodon - Collegamento all'originale
Today's featured color from Storied Colors: storiedcolors.com/color/imperi…

Imperial Yellow, Storied Colors

The exclusive yellow of Chinese emperors from the Tang dynasty onward, forbidden by sumptuary law to anyone outside the imperial family.
Storied Colors

reshared this

Lorenzo ha ricondiviso questo.

mastodon - Collegamento all'originale
A “model” is a large matrix of weights to infer outputs given a set of inputs. It can not break out of a sandbox unless equipped with an execution environment intentionally weaponised to provide it hostile execution capability. But because it is a weighted random number generator, it will be able to iterate through all inputs and outputs to find all combinations of potential results.
nytimes.com/2026/07/21/technol…

reshared this

in reply to Osma A 🇫🇮🇺🇦

mastodon - Collegamento all'originale

Osma A 🇫🇮🇺🇦

Somehow, we have ended up in a universe where companies brag about deploying multibillion dollar software systems with an objective to break decades old IT security legislation, and nobody holds their executives to account.
wired.com/story/anthropic-says…

Anthropic Says Claude Hacked 3 Organizations During Cybersecurity Tests

In a review triggered by OpenAI’s Hugging Face incident, Anthropic discovered three of its AI models had breached real organizations during third-party evaluations.
Louise Matsakis (WIRED)
Lorenzo ha ricondiviso questo.

mastodon - Collegamento all'originale
The media in this post is not displayed to visitors. To view it, please go to the original post.

-Non-profit offers $22,000 bounty for INC ransomware group
-Hackers breach the UK Department for Education
-Russia charges Pavel Durov
-FCC bans foreign robots and power inverters
-Minnesota water attacks linked to Iran
-Adform hacked to spread malware
-OpenAI-Hugging Face come out
-Unitel hacked ahead of IPO
-SplitVPN hacked
-CubePilot DNS hijacking event
-KT fined $37m for breaches
-CareCloud breach
-ExfilSquad goes on a hacking spree

N: news.risky.biz/risky-bulletin-…
P: risky.biz/RBNEWS594/

Non-profit offers $22,000 bounty for INC ransomware group

In other news: Hackers breach the UK Department for Education; Russia charges Pavel Durov; FCC bans foreign robots and power inverters.
Catalin Cimpanu (Risky.Biz)

reshared this

in reply to Catalin Cimpanu

mastodon - Collegamento all'originale

Catalin Cimpanu

The media in this post is not displayed to visitors. To view it, please go to the original post.
-New GenieLocker ransomware
-LogoKit and ARToken PhaaS
-Lots of DPRK APT reports
-Laundry Bear linked to recent Exchange zero-day
-New Cisco zero-day
-MikroTik brute-force bug
-New Copilot Word worm
-CosmosEscape vulnerability impacts Azure's CosmoDB
-Vulns in Volvo's fleet portal
-Okta buys Permiso Security
-DEFCON bans smart glasses
-New SBOM guidance
-OpenAI releases Codex Security CLI
in reply to Catalin Cimpanu

mastodon - Collegamento all'originale

Gilgwath

yes, the LLM that on the 24th wishes me good luck for tomorrow, when that thing is on the 28th. Repeatedly, after I corrected it twice. That LLM cracked the some of the hardest math of the century? Sure! Together with the full squeezing of a saskatchewan DNA sample.

What's next? CAD drawings for a negative energy source to put into a warp drive? They must run out of outrageous claims at some point, right?

Lorenzo ha ricondiviso questo.

mastodon - Collegamento all'originale

Idiots.

“US government lists fictional nation Wakanda as trade partner”

bbc.com/news/world-us-canada-5…

US government lists fictional nation Wakanda as trade partner

Officials say it was an error to add the fictional nation from Black Panther to the list.
BBC News
Lorenzo ha ricondiviso questo.

mastodon - Collegamento all'originale
Adform, a web advertising firm, have been hacked and have been serving a supply chain attack to steal crypto doublepulsar.com/adform-compro…

Adform compromised to serve crypto stealer via supply chain attack

Large web advertising firm serving crypto wallet stealer and C2 beacon traffic.
Kevin Beaumont (Medium)
Lorenzo ha ricondiviso questo.

mastodon - Collegamento all'originale
The media in this post is not displayed to visitors. To view it, please go to the original post.

Bruh... there's a "master key" that grants access to every Cosmos DB on Azure? Wut?

wiz.io/blog/cosmosescape-takin…

CosmosEscape: Taking Over Every Azure Cosmos DB | Wiz Blog

Wiz Research details CosmosEscape, a critical vulnerability in Azure Cosmos DB that granted full read/write access to every database. Now fully remediated.
Yuval Avrahami (wiz.io)
in reply to Catalin Cimpanu

mastodon - Collegamento all'originale

Jan Katins

every DB at least needs a place where the internal key to setup replicas is placed. I guess every DB vendor has such a key saved (per database)bin their control plane data store. E.g. AWS only gives you a non superuser credential (or gave you a few years back) and I guess the real superuser was reserved for internal stuff like replica setup and monitoring.
Lorenzo ha ricondiviso questo.

mastodon - Collegamento all'originale

Turkish messenger Bip and South Korean messenger KakaoTalk have stopped working in Russia about the same time authorities charged Durov.

Something, something, ban all IM services until citizens use the state-owned MAX, bla bla

currenttime.tv/a/problema-v-ra…

Пользователи сообщили о проблемах в работе приложений KakaoTalk и BiP, которые в России стали заменой заблокированным мессенджерам

Российские пользователи корейского KakaoTalk и турецкого BiP сообщили "Осторожно, Москва", что уже два дня эти мессенджеры плохо работают, если не включен VPN.
Настоящее Время

reshared this

Lorenzo ha ricondiviso questo.

mastodon - Collegamento all'originale

📺 Srsly Risky Biz: Chipping away at Chinese AI risks

risky.biz/video/srsly-risky-bi…

Srsly Risky Biz: Chipping away at Chinese AI risks - Risky Business Media

Tom Uren and James Wilson talk about open-weight AI models and distillation. These topics have been subject to a lot of US government atte [Read More]
risky.biz

reshared this

⇧