Originally known as FORTRAN, but written in lower case since the 1990s with Fortran 90, this language was developed initially by John Backus as a way to make writing programs for the IBM 704 mainframe easier. The 704 was a 1954 mainframe with the honor of being the first mass-produced computer that supported hardware-based floating point calculations. This functionality opened it up to a whole new dimension of scientific computing, with use by Bell Labs, US national laboratories, NACA (later NASA), and many universities.

Much of this work involved turning equations for fluid dynamics and similar into programs that could be run on mainframes like the 704. This translating of formulas used to be done tediously in assembly languages before Backus’ Formula Translator (FORTRAN) was introduced to remove most of this tedium. With it, engineers and physicists could focus on doing their work and generating results rather than deal with the minutiae of assembly code. Decades later, this is still what Fortran is used for today, as a domain-specific language (DSL) for scientific computing and related fields.

In this introduction to Fortran 90 and its later updates we will be looking at what exactly it is that makes Fortran still such a good choice today, as well as how to get started with it.

Modern Fortran

Punch card from a typical FORTRAN program by the early 1970s. (Credit: Arnold Reinhold, Wikimedia)
The release of the Fortran 90 (F90) specification in 1991 was the first major update to the language since Fortran 77, and introduced many usability improvements, as well as the dropping of punch card era legacy requirements and limitations to variable lengths and more. This is the reason for our focus on F90 here, as there is no real reason to use F77 or earlier, unless you’re maintaining a legacy codebase, or you have a stack of new cards that need punching. In case you are dying to know what changed, the Wikibooks Fortran Examples article has examples of early FORTRAN all the way to modern Fortran.

Of note here is that a modern Fortran compiler like GCC’s GFortran(forked from g95) still supports F77, but users are highly encouraged to move on to Fortran 95, a minor update to F90, with GFortran supporting up to F2008, with coarray support, as covered later. F2018 support is still a work in progress as of writing, but many features are already available.

Support for the latest standard (F2023) is not widely available yet outside of commercial compilers, but as a minor extension of F2018 it should eventually get rolled into those features as the implementation progresses. This means that for now F2008 is the latest standard we can reliably target across toolchains with new Fortran code.

Beyond GFortran there are a few more options, including Flang in LLVM and LFortran in addition to a gaggle of commercial offerings. Unless you intend to run high-performance computing code on massive parallel clusters like supercomputers, the GNU and LLVM offerings will probably suffice. Simply fetch either GFortran or Flang from your local package manager or equivalent and you should be ready to start with programming in Fortran.

Hello World

As with most DSLs, there is very little preamble to start writing the business logic. The ‘Hello World’ example is most succinct:
program helloworld
print *, "Hello, World!"
end program helloworld
The program name is specified right after the opening program keyword, which is repeated after the closing end program. This is similar to languages like Ada and Pascal. The program name does not have to match the name of the file. Although there’s no explicit specification for what the file extension has to be for a Fortran source file, convention dictates that for F77 and older you use .f or .for, while F90 and newer uses generally .f90 as extension. Although some opt to use extensions like .f95 and .f03, this is rather confusing, isn’t recognized by all compilers and all of those are similar free-form Fortran source files anyway.

Tl;dr: Use .f90 for modern Fortran source files. Our Hello World example goes into a file called hello_world.f90.

The other point of note in this basic example is the print command, which looks somewhat cryptic but is quite easy. The first argument is the format, reminiscent of C’s printf. The asterisk here simply means that we use the default format for the provided value, but we could for example print the first string as an 11 character wide field and a variable string as 8 wide:
character(len=8) :: name = 'Karl'
print '(a11,a8)', 'My name is ', name
This also shows how to declare and define a variable in Fortran. Note that if you do not start the code with implicit none, variable names that start with I through N are considered to be integer type and real otherwise. With gfortran you can also globally do this by compiling with the -fimplicit-none flag.

A total of five basic types are supported:

• real
• integer
• logical (boolean)
• complex
• character

Finally, comments in Fortran are preceded by an exclamation mark !. It’s also relevant to note that Fortran – like all good programming languages – is case insensitive, so you can still write your Fortran code like it’s F77 or Fortran II, yelling in all caps without anyone but the people reading your code batting an eye.

Hello Science

Now that we have got a handle on the basics of Fortran, we can look at some fun stuff that Fortran makes really easy. Perhaps unsurprisingly, as a DSL that targets scientific computing, much of this fun stuff focuses around making such types of computing as easy as possible. Much of this can be found in the intrinsic procedures of Fortran, which make working with real, integer and complex values quite straightforward.

For example, conjugating a complex number with conjg:
(Credit: Fortran Wiki)
Basically, whatever mathematical operation you wish to perform, Fortran should have you covered, allowing you to translate your formulas into a functional program without having to bother with any dependencies or the like. This includes working with matrices and getting into the weeds with numerical precision.

Even better is that you’re not stuck running your code on a single CPU core either. Since Fortran 2008, Coarray Fortran (CAF) is now part of the specification. This feature enables parallel processing, which is generally implementing on top of the Message Passing Interface (MPI) protocol, with gfortran implementing CAF support. Depending on the selected option with the -fcoarray= flag, gfortran can use the ‘single’ image (thread) option, or with a library like OpenCoarrays it can use MPI, GASNet, and others.

When using CAF with MPI, the program (‘image’) is distributed across all nodes in the MPI cluster per its configuration, with synchronization occurring as defined by the program. With OpenCoarrays available from many local OS repositories, this means that any budding molecular scientist and astrophysicists can set up their own MPI cluster and start running simulations with relatively very little effort.

The DSL Life

Much like when we looked at COBOL, a DSL like Fortran is often misunderstood as ‘yet another programming language’, much like how in the 1980s some thought that the scientific and engineering communities were going to switch over to Pascal or Modula-2. One simple reason that didn’t happen lies in the very nature of DSLs, with them being developed explicitly to deal with that specific domain. It will always be possible to do everything a DSL does in any generic programming language, it’s just that a DSL can be optimized more exactly, and is often easier to maintain, since it is not generic.

There are many things that Fortran does not have to concern itself with, yet which haunt languages like C and its kin. Meanwhile, reimplementing Fortran in C would come at considerable cost, run into certain limitations of the language and potentially require compromises in order to get close to the original functionality of the DSL.

Running a cluster with Coarray-based Fortran source, churning through complex simulations that require utmost control over precision and where one small flaw can waste days of very expensive calculations, those are the kind of scenarios where you’re not looking for some generic language to poorly reinvent the wheel, but where you keep using the same – yet much more refined – wheel that has gotten us this far.

hackaday.com/2025/08/04/a-gent…

Steve and James discuss ICE (et. al.) actions, how they are becoming a secret police force and why we need to get rid of them.

youtube.com/embed/SM6kGmzkUFA?…

Sign up to our newsletter to get notified of new events or volunteer. Join us on:

• Mastodon;
• Facebook;
• Blue Sky;
• Twitter.

Some links we mentioned:

• MassPirates 2025 Administrative Coup Memory Bank;
• The IRS Is Building a Vast System to Share Millions of Taxpayers’ Data With ICE;
• 26 USC 6103: Confidentiality and disclosure of returns and return information;
• Immigration agents demand tenant information from landlords, stirring questions and confusion;
• Flight Manifests Reveal Dozens of Previously Unknown People on Three Deportation Flights to El Salvador;

Image Credit: Public Domain via Wikipedia Commons

masspirates.org/blog/2025/08/0…

Three-axis 3D printing has been with us long enough that everybody knows the limitations, but so far, adding extra axes has been very much a niche endeavor. [Daniel] at Fractal Robotics wants to change that, with the Fractal 5 Pro 5-axis printer, and its corresponding Fractal Cortex slicer.

The printer looks like an extra-beefy Voron from a distance, which is no surprise as [Daniel] admits to taking heavy inspiration from the Voron Trident. The Fractal 5 shares a core-XY geometry with the Voron, using beefy 30 mm x 30 mm extrusions. Also like the Voron, it runs Klipper on a Raspberry Pi hiding in the base. Under a standard-looking printhead using a BondTech extruder and E3D volcano hotend, we find the extra two axes hiding under the circular build plate. The B axis is a gantry that can pivot the build plate assembly a full 90 degrees; the A axis spins the plate without limit thanks to the slip rings built into the design.

The extruder may look fairly normal, but it has actually been designed very carefully to allow the nozzle to get as close as possible to the build plate when the B-axis is at 90 degrees. It looks like the E3D hotend is actually the limiting factor there, which gives plenty of design freedom when planning prints in the accompanying Fractal Cortex slicer.

Fractal Cortex is an all-new slicer written entirely in Python. It does have a 3-axis legacy mode, but it’s primarily designed for 5-axis slicing with the Fractal 5 Pro. The multi-axis operation looks very straightforward: you place “slicing planes” and orient them along the part, as many as you require. Printing pauses while the A and B axes rotate, then resumes with straight, parallel layers. Sure, non-planar slicing would be the bees’ knees on this sort of 5-axis printer, but we’ve got to say for a young engineer’s first crack at this kind of software, Fractal Cortex looks pretty good.

It sounds like [Daniel] is looking for contributors to the project, so if this project tickles your fancy, head over to the GitHub links at the top of the page and dive in.

We’ve seen other 5D printers before, but this one probably takes the cake for build volume, and having a slicer to match is a big advantage for anyone who wants to try this at home.

Thanks to [Hari] for the tip. Use all your axes to extrude your projects into our tips line here.

youtube.com/embed/2_BY-7XOgWg?…

hackaday.com/2025/08/04/open-s…

The MOS Technology 6502 is a microprocessor which casts a long shadow over the world of computing. Many of you will know it as the beating heart of so many famous 8-bit machines from the likes of Commodore, Apple, Acorn, and more, and it has retained enough success for a version to remain in production today. It’s still a surprise though, to note that this part is now fifty years old. Though there are several contenders for its birthday, the first adverts for it were in print by July 1975, and the first customers bought their chips in September of that year. It’s thus only fitting that in August 2025, we give this processor a retrospective.

The Moment Motorola Never Really Recovered From

The advert that started it all. MOS Technology, Public domain.
The story of the 6502’s conception is a fascinating tale of how the giants of the early mocroprocessor industry set about grappling with these new machines. In the earlier half of the 1970s, Chuck Peddle worked for Motorola, whose 6800 microprocessor reached the market in 1974. The 6800 was for its time complex, expensive, and difficult to manufacture, and Peddle’s response to this was a far simpler device with a slimmed-down instruction set that his contact with customers had convinced him the market was looking for: the 6502.

There’s a tale of Motorola officially ordering him to stop working on this idea, something he would later assert as such an abandonment of the technology that he could claim the IP for himself. Accompanied by a group of his Motorola 6800 colleagues, in the summer of 1974 he jumped ship for MOS Technology to pursue the design. What first emerged was the 6501, a chip pin-compatible with the 6800, followed soon after by the 6502, with the same core, but with an on-board clock oscillator.
My 6502-powered Beeb may be showing its age, but it remains one of the most comprehensively specified computers for its time I have ever used.
If you were one of the small group of very geeky kids in the early 1980s who really used their home computers instead of merely gaming it’s likely you are part of the only generation ever to have a playground one-upmanship over processor architectures. Those of us with a Z80 in our Sinclair machines would cite the higher clock speed, while the Commodore owners would talk about the 6502’s pipelined architecture and smaller instruction set.

With the benefit of many decades’ hindsight it’s fair to say they had a point, and one that the Z80’s ability to flip between two sets of registers on a dime couldn’t make up for. The 6502 was quick, it had a reliable manufacturing process, and since MOS Technology sold it to anyone for a surprisingly reasonable price rather than restricting it to big-bucks industrial customers, it or one of its many derivatives was a natural choice for much of the first generation of home computers and consoles. If you had an Acorn, an Atari, a Commodore, an Oric, or an Apple 8-bit machine you would have had one, and if you had a Nintendo console you also joined the 6502 club.

Just Who Is Still Buying A 6502?

It’s still not too difficult to find dedicated defenders of the 6502 among those now in their 50s geeks who owned an 8-bit computer when they were new. Just as with their Z80-fan friends, they come from an era when it was possible to fully comprehend the architecture of your computer at nearly gate level. It’s thus understandable that both architectures lingered on in the world of embedded computing long after their heyday on the desktop had passed, but it’s still a surprise to find the WDC version of the chip still in production. Unlike other surviving 8-bit architectures such as the 8051, this is the real thing in a 40-pin DIP rather than a core found embedded in other chips, so we’d be curious to know: aside from a small number of retrocomputing enthusiasts, who is still buying these things? The Z80 finally shuffled off this mortal coil last year, so how long has the WDC W65C02 got left? We’d be curious for your views.

Speaking personally, I had a foot in both camps. Like many British kids of the early 1980s I could save enough pocket money for a Z80-based Sinclair ZX81 of my own, but I had access to a room full of 6502-powered Acorn BBC Micros at school. I thus painstakingly hand-assembled Z80 code for the Sinclair, and was blown away by the built-in assembler in the Beeb. But did I have a favourite? Perhaps not, after all the most important lesson is one I only fully understood years later; that far more important than what architecture a computer uses is what you can do with it. And in that, to my profound sadness as a Sinclair fan, the 6502-powered BBC Micro won hands-down.

Header image: Dirk Oppelt, CC BY-SA 3.0.

hackaday.com/2025/08/04/happy-…

Attraverso la funzionalità Component Object Model (COM) di BitLocker, gli aggressori possono mettere in atto una tecnica innovativa di pivoting, finalizzata all’esecuzione di codice malevolo sui sistemi presi di mira. Questo approccio, presentato tramite uno strumento di prova chiamato BitLockMove, segna un progresso significativo nelle strategie di movimento laterale, riuscendo a eludere i metodi di individuazione standard grazie all’utilizzo di elementi Windows autentici.

Solitamente abilitato su workstation e laptop per impedire l’accesso non autorizzato in caso di furto o smarrimento del dispositivo, la protezione completa di BitLocker lo ha reso un bersaglio allettante per gli aggressori che cercano di abusare della sua infrastruttura sottostante.
Lateral Movement – BitLocker Diagram (fonte ipurple.team)
Durante la sua presentazione, il ricercatore Fabian Mosch ha illustrato che l’aggiunta di ogni applicazione o funzionalità a Windows comporta un aumento sostanziale degli oggetti, comprendenti processi, file e chiavi di registro, i quali, nel loro insieme, ampliano la superficie esposta a potenziali attacchi. Nonostante l’efficacia di BitLocker nella protezione dei dati non attivi, gli esperti hanno rilevato che la sua implementazione include aspetti che possono essere sfruttati dai malintenzionati più esperti per creare strumenti offensivi.

La tecnica appena divulgata mira alla manipolazione remota delle chiavi del registro di sistema di BitLocker tramite Windows Management Instrumentation (WMI) per dirottare specifici oggetti COM. Questo approccio consente agli aggressori di eseguire codice nel contesto dell’utente interattivo sugli host di destinazione, portando potenzialmente all’escalation del dominio se l’utente compromesso possiede privilegi elevati, come i diritti di amministratore di dominio.

Una falla critica nella struttura gerarchica degli oggetti COM di BitLocker è sfruttata dall’attacco, che punta direttamente alla classe BDEUILauncher attraverso l’interfaccia IBDEUILauncher. Gli aggressori possono utilizzare tre metodi principali offerti da questa interfaccia per il loro sfruttamento:

• BdeUIProcessStart : avvia il processo BitLocker (BdeUISrv.exe)
• BdeUIContextTrigger : fornisce funzionalità di manipolazione del contesto
• GetUserLogonTime : recupera le informazioni sui tempi di accesso dell’utente

Il processo di exploit si concentra sul CLSID ab93b6f1-be76-4185-a488-a9001b105b94, che genera quattro processi diversi come utente interattivo. Tra questi, il processo BaaUpdate.exe si dimostra particolarmente vulnerabile al COM hijacking quando eseguito con parametri di input. L’attacco prende di mira specificamente il CLSID mancante A7A63E5C-3877-4840-8727-C1EA9D7A4D50, che il processo BaaUpdate.exe tenta di caricare.

Creando una voce di registro per questo CLSID e stabilendo le sottochiavi appropriate, gli aggressori possono reindirizzare il processo per caricare codice dannoso invece del componente legittimo, ha affermato Fabin . Lo strumento BitLockMove dimostra l’implementazione pratica di questa tecnica attraverso due modalità operative:

• Modalità di enumerazione: La funzionalità di ricognizione dello strumento utilizza API Microsoft non documentate della libreria winsta.dll per enumerare da remoto le sessioni attive sui sistemi di destinazione. Queste API, tra cui WinStationEnumerateW, WinStationOpenServerW e WinStationQueryInformationW, forniscono informazioni complete sulle sessioni senza richiedere l’abilitazione dei Servizi Desktop remoto.
• Modalità di attacco: Durante la fase di exploit attivo, BitLockMove stabilisce una connessione remota all’host di destinazione tramite WMI ed esegue query per abilitare il servizio Registro di sistema remoto. Lo strumento crea quindi il percorso di registro necessario per preparare l’ambiente al dirottamento COM, creando in particolare voci nella struttura di chiavi CLSID.

BitLockMove – Enumerazione delle sessioni remote (fonte ipurple.team)
Il processo BdeUISrv.exe rimane attivo sul sistema, a differenza del processo vulnerabile al COM hijacking BaaUpdate.exe che viene terminato.

È importante notare che il processo BdeUISrv.exe è stato eseguito nel contesto dell’utente Amministratore, ovvero l’utente che aveva la sessione interattiva sull’host ed è stato preso di mira.

Sebbene la tecnica tenti di integrarsi nell’ambiente, poiché il comando arbitrario viene eseguito nel contesto di un processo BitLocker attendibile (BdeUISrv.exe ), esistono molteplici opportunità di rilevamento in varie fasi. L’analisi del comportamento dello strumento può rivelare le caratteristiche della tecnica e le aree su cui gli ingegneri del rilevamento dovrebbero concentrarsi.

La tecnica di spostamento laterale tramite dirottamento del modello a oggetti dei componenti BitLocker prevede diverse fasi, ma offre anche numerose opportunità di rilevamento. Le organizzazioni dovrebbero valutare quali log sia possibile abilitare nei propri ambienti e concentrare di conseguenza i propri sforzi di ingegneria del rilevamento.

L’arricchimento del SIEM con tutti i log associati e l’esecuzione di query di threat hunting a breve frequenza consentono il rilevamento con un elevato livello di affidabilità.

L'articolo BitLocker sotto Attacco! Una nuova tecnica consente movimenti laterali eludendo i controlli proviene da il blog della sicurezza informatica.

IT'S MONDAY, AND THIS IS DIGITAL POLITICS. I'm Mark Scott, and here's a little plug for my day job.

As part of work I'm doing linked to the United Kingdom's efforts around social media transparency and accountability, we'll be holding a webinar with regulators and policymakers on Aug. 6 (at 11am UK time / 6am ET — sorry!) about the country's new data access efforts. You can sign up here.

— Google and Meta's collective pullback on political ad transparency in the European Union harms free speech and the public's engagement with political campaigns.

— What Big Tech's recent earnings season demonstrates about their high-wire geopolitical balance between the United States and everyone else.

— Almost half of the United Kingdom has never used artificial intelligence tools in their daily lives.

Let's get started:

THE END OF DIGITAL POLITICAL ADS

LATE LAST YEAR, GOOGLE PUT OUT A POST that few outside of digital policy circles paid much attention to. In the 2-minute read, the search giant's head of EU government affairs said that, as of October, 2025, the tech giant would not allow advertisers to buy political ads targeted within the 27-country bloc. The reason? An obscure new rule, known as the EU Regulation on Transparency and Targeting of Political Advertising, or TTPA.

The rule "introduces significant new operational challenges and legal uncertainties for political advertisers and platforms," read the announcement. "Google will stop serving political advertising in the EU before the TTPA enters into force in October 2025."

Then, on July 25, Meta followed suit. In an equally short and obscure message, the social media company said it too would be ending its political ad offering inside the EU because of the upcoming TTPA. "We continue to believe online political advertising is a vital part of modern politics," Meta said in its post. "The TTPA introduces significant, additional obligations to our processes and systems that create an untenable level of complexity and legal uncertainty for advertisers and platforms operating in the EU."

And with that, the two largest platforms for political ads within the second largest democratic bloc (behind India) hobbled politicians' ability to talk to would-be supporters. On top of that, Google and Meta made it almost impossible to track what was being spent on political campaigns, by who, and who was the audience for these paid-for messages.

Thanks for reading the free monthly version of Digital Politics. Paid subscribers receive at least one newsletter a week. If that sounds like your jam, please sign up here.

Here's what paid subscribers read in July:
— We are entering a new era of social media where everyone lives in digital siloes and the cost of spreading falsehoods is next to nothing. More here.
— The EU-US trade dispute will not include digital; Where the EU and US diverge on AI governance; Tech-related risks now top people's concerns across multiple different regions worldwide. More here.
— The rise of 'age verification' across the West is ending the era of online anonymity; All you need to know about the White House's upcoming AI Action Plan; The billions to be spent on digital in the next EU Budget. More here.
— The US-EU trade/tariff agreement was a skirmish in a wider tech-focused war; Contrasting American and Chinese AI strategies; Energy consumption linked to AI, data centers and crypto will double between 2022-26. More here.

It was a one-two punch that both hit people's legitimate free speech rights and hamstrung wider society's obligations to understand how national elections were influenced, by domestic and foreign groups.

First, a quick synopsis of the (flawed) TTPA. Introduced in 2024 with the other alphabet soup of EU digital regulation, the rules were aimed at giving citizens greater awareness about the types of political ads they saw online.

That included 1) ensuring political ads were labeled as such; 2) accessible information, stored in an online repository, about who paid for each digital ad; 3) a ban on foreign entities paying for these paid-for messages in the build-up to national elections; and — most importantly — 4) new restrictions that limited platforms and advertisers in targeting people via data the firms/political groups had already collected on individuals unless they explicitly consented to receiving political ads.

The new limits on how political ads could be targeted was a major stumbling block. It made it next to impossible for Meta and Google to help political groups to target would-be supporters. To do that, the companies and/or politicians would have to explicitly ask people if they were OK about receiving digital political ads — something that even the most ardent political supporter would likely be beige about, at best.

There are legitimate weaknesses in the TTPA that everyone knew about from the beginning.

The definition of "political ads" varied widely between platforms, and EU policymakers' efforts to harmonize that — alongside so-called "issues-based" ads like those around climate change, public health and other societal issues — spread the net too wide. In practice, that meant a large bucket of ads beyond traditional party political messaging would likely be caught within the TTPA. That, in turn, would hit platforms' underlying advertising business models with the potential of 6 percent fines (linked to a company's yearly revenue) for failing to comply.

Faced with that regulatory uncertainty — and at a time when Google and Meta are trying (badly) to pull back from politics — it made sense both firms would give up a part of their business that was marginal to their collective bottom lines. Why take the risk if you could just pull the plug in a way that, in theory, would anger some politicians already falling out of love with Europe's penchant for digital regulation?

And yet, I just don't buy in.

Negotiations were still underway about how the TTPA would be implemented, as of October, so pulling the plug now was an active choice from both Google (in November, 2024) and Meta (in July.) The companies would argue that, at some point, they had to call it a day. And that was their choice. But after companies doubled down on their free speech bonafides, the decision to then limit people's free speech rights via buying political ads felt out of step. The separate decisions would be particularly hard for smaller political movements that could level the playing field with larger rivals via targeted digital political ads.

I also don't believe either company doesn't have the internal resources (both technical and financial) to find a solution. Yes, the definition of what constitutes a "political ad" in the TTPA is bad. But both firms have been tracking various types of ads for years. To suggest that engineers — or even artificial intelligence tooling — could not figure out the line between what was allowed, and what was not, also felt too binary.

When you then look at how much these firms are making (more on that in the section below) on a quarterly basis, this looks more like a political decision than an economic one. Alphabet (Google's parent company), for instance, plans to spend $86 billion this year, primarily on AI-related infrastructure. That's on top of the $28 billion net profit the company made in the second quarter of 2025.

When spending/making that amount of money, the idea that figuring out how to implement (flawed) political ad regulations was too costly doesn't add up.

There is still time, before the October deadline, for the tech companies and the European Commission to find a solution. But that's not really how this should work. It is Meta and Google's right to pull back on services within the bloc — even if sometimes that feels more like point-scoring than actual policymaking.

But without a coherent approach to overseeing political ads in one of the world's largest democratic blocs, these companies are leaving a massive hole in which nefarious actors will likely continue peppering voters with political ads — but in a way that lacks transparency, accountability and oversight. That will harm legitimate political actors who will no longer have a legal way, within the bloc, to amplify their messaging to would-be supporters.

Faced with that trade-off, both companies have made the wrong choice. Google and Meta will still play a central role in domestic European politics, even if they follow through with their respective decisions to pull out of the political ad business in Europe.

By abdicating this responsibility, the firms are not, collectively, living up to their stated objectives (to wider society) of uplifting people's free speech rights in a way that promotes transparency and accountability for all.

Chart of the Week

ROUGHLY 40 PERCENT OF BRITS never use artificial intelligence tools as part of their daily lives, and only 10 percent of those polled by YouGov say they use it at least once or twice a day.

For those who do rely on the emerging technology, they most regularly seek information about "a simple topic." The least used reason is to find out what is going on in the news.

Source: YouGov

BIG TECH EARNINGS, AI AND GEOPOLITICS

AS A LAPSED FINANCE JOURNALIST, I can't help but scan tech companies' earnings reports when they are published each quarter. Policymakers and politicians may not always tell the truth. But numbers don't lie, so these regular updates from Silicon Valley's biggest names offer a way to cut through the noise to understand how these firms are making and spending their money.

The figures are wild. Meta's net income in the second quarter, or the profit that the social media giant made after taxes, was a whopping $13.5 billion, or a 36 percent yearly jump. Microsoft's figure rose 22 percent, to $27.2 billion, over the same period. Amazon — not a company known for making much profit, despite its size — recorded net income for the three months through June 30 or $18.2 billion, or a 34% increase.

These profits are crazy. But that's not where the politics lie in the companies' separate announcements.

Instead, the combination of Google, Amazon, Meta and Microsoft said that, during the 2025-26 fiscal year, they had earmarked roughly $370 billion for spending, primarily to drive big bets on artificial intelligence. That includes massive spending on data centers, AI engineers and other AI-related costs required in what is increasingly turning into a Silicon Valley-led race toward AI dominance.

Let's put that into perspective.

The $370 billion back-of-the-napkin figure doesn't include spending from the second tier of AI (Western) giants, so it's an inherently conservative number. It also doesn't include other digital infrastructure spending, tax cuts and research and design expenditure already announced by the likes of the United States and EU. At best, it's a significant under-count on what has already been allocated by these firms — and that comes despite many of the companies recording significant increases in their net profits (see above.)

Sign up for Digital Politics

Thanks for getting this far. Enjoyed what you've read? Why not receive weekly updates on how the worlds of technology and politics are colliding like never before. The first two weeks of any paid subscription are free.

Subscribe
Email sent! Check your inbox to complete your signup.

No spam. Unsubscribe anytime.

Here's another stat: The $370 billion figure — solely linked to Big Tech capital expenditure — is the equivalent to the expected 2025 gross domestic product of the Czech Republic, or the world's 41st largest GDP. In anyone's mind, that is a lot of money.

Here's where the geopolitics comes in.

These firms are American — and some have lent heavily into the Red, White and Blue to win favor with the current White House administration. But much of what has been earmarked in AI-related spending is destined for outside of the US. That includes significant expansion in companies' global networks of data centers to meet non-US needs. In those conversations, the companies have often played down their American roots, highlighting how they can meet the domestic objectives of national politicians eager to develop indigenous AI tooling to compete on the global stage.

In this bizarro world, two opposite, but complementary, messages are playing out, in real time.

For US policymakers, these firms are signed up to the American "dominance" agenda, as articulated in the White House's recent AI Action Plan. For non-US policymakers, the companies are pushing a "we're here to help" ethos that positions them as apolitical globalist allies that can neutrally support these countries' often contradictory policy aims to those of the US.

Central to those conversations are the megabucks announced in the latest earnings cycle. The expected capital expenditure is the realization of these political discussions, both at home and abroad, at a time when all of Silicon Valley is desperately seeking an edge to keep competitors at bay. That means walking a tight rope between American domestic interests and the business opportunities overseas.

What many in Washington have yet to understand is how these conversations are playing out in national capitals from Brasilia to Berlin. Once American policymakers figure that out, it's an open question if they will view the companies' international outreach as either part of American "dominance" or as a potential direct conflict with the White House's MAGA agenda.

What I'm reading

— UK Research and Innovation, a public research body, announced a $74 million program to attract world-class researchers to the UK. More here.

— An Australian court ruled that X would have to respond to a local regulator's requests on how the social network tackled child sexual abuse material. More here.

— Michael Kratsios, the director of The White House's Office for Science and Technology Policy, spoke at the Center for Strategic & International Studies about the AI Action Plan. More here.

— Donald Trump's administration announced tariffs against Brazil, in part related to how the South American country approached questions around content moderation and the role of platforms in the failed 2022 coup. More here.

— Italy's competition regulator announced an investigation into Meta's potential abuse of its market position by installing an AI agent within its WhatsApp service. More here.

digitalpolitics.co/newsletter0…

Mozilla ha avvisato gli sviluppatori di estensioni per Firefox di una nuova campagna di phishing volta a compromettere i loro account sulla piattaforma ufficiale AMO (addons.mozilla.org). Questo ecosistema include oltre 60.000 componenti aggiuntivi e più di mezzo milione di temi, utilizzati quotidianamente da decine di milioni di persone in tutto il mondo.

Secondo l’avviso pubblicato, gli aggressori inviano email per conto del team AMO, sostenendo che l’account sviluppatore deve essere aggiornato urgentemente per mantenere l’accesso agli strumenti. In realtà, tali email conducono a siti falsi creati allo scopo di rubare login e password. Il messaggio contiene solitamente una variante della dicitura “Il tuo account dei componenti aggiuntivi Mozilla richiede un aggiornamento per continuare a utilizzare le funzionalità per sviluppatori”, che ha lo scopo di allarmare il destinatario e indurlo a cliccare sul link dannoso.

Mozilla raccomanda vivamente di verificare l’autenticità delle email: devono provenire dai domini firefox.com, mozilla.org, mozilla.com o dai loro sottodomini, e devono essere sottoposte ad autenticazione di base tramite SPF, DKIM e DMARC. Si consiglia agli sviluppatori di evitare di cliccare sui link presenti in tali messaggi e, se necessario, di visitare autonomamente il sito web di AMO tramite l’indirizzo ufficiale per assicurarsi che le informazioni siano aggiornate. Si sottolinea in particolare che login e password devono essere inseriti solo sui siti web originali di Mozilla o Firefox.

Sebbene la portata dell’attacco non sia ancora stata rivelata, Mozilla ha confermato che almeno uno sviluppatore è già caduto vittima del sistema. Ciò suggerisce una minaccia reale, sebbene non siano ancora disponibili dati sul numero di account compromessi o sulle prossime mosse degli aggressori. L’organizzazione ha promesso di fornire ulteriori informazioni non appena disponibili.

Questa situazione solleva un contesto importante. Il mese scorso, il team Add-ons Operations ha implementato una nuova misura di sicurezza progettata per bloccare automaticamente le estensioni dannose mascherate da wallet di criptovalute. Come ha osservato il responsabile del team, Andreas Wagner, negli ultimi anni sono state identificate e rimosse centinaia di estensioni pericolose. Alcune di queste venivano utilizzate direttamente per rubare criptovalute, sebbene non tutte mostrassero evidenti segni di danno.

In questo contesto, le statistiche sono particolarmente allarmanti: solo lo scorso anno, i criminali sono riusciti a rubare circa 494 milioni di dollari in criptovalute attaccando wallet, inclusi oltre 300.000 indirizzi univoci. Casi come questi dimostrano quanto possa essere pericoloso anche un singolo attacco hacker all’account di uno sviluppatore: data la diffusione capillare dei componenti aggiuntivi, questi diventano una piattaforma ideale per l’introduzione di codice dannoso.

La conclusione in breve è questa: gli sviluppatori che pubblicano estensioni sulla piattaforma Mozilla sono di nuovo a rischio. Il phishing in questo caso non è solo una minaccia per la sicurezza personale, ma un potenziale canale per infettare decine di milioni di utenti in tutto il mondo.

L'articolo Mozilla avverte gli sviluppatori di estensioni Firefox: campagna di phishing attiva proviene da il blog della sicurezza informatica.