The News,
Decoded.

Oh, marvellous. Maryland has decided that if the federal government can deploy AI with all the transparency of a Victorian séance, why not join the party? More than 50 state agencies are now using AI tools, according to a public inventory that probably reads like the minutes of a meeting that never happened. Questions about oversight? Naturally. But who needs oversight when you have a spreadsheet that says 'We're using AI, trust us'. This is the same glorious opacity we flagged across federal agencies, now trickling down to state level.

The inventory is a masterclass in plausible deniability: lists tools, but good luck figuring out what they actually do or what data they chew through. One imagines the state's IT department filled out the form with vague entries like 'chatbot for things' and 'automated decision-making on unspecified matters'. The 'more than 50' figure is almost certainly a floor, not a ceiling — counting AI deployments from agencies that can barely keep their websites running without a governor's executive order is a fool's errand.

So here we are: the great AI rollout continues, with the same old promise of efficiency and the same old refusal to say how it actually works or who gets hurt when it breaks. Maryland's citizens can rest easy knowing their tax dollars are funding systems their own government can't be bothered to explain. Raise a glass of Old Bay-infused resignation. At least the algorithm won't care about your crabs.

Oh, what a delightful twist: the very labs racing to build models that could theoretically help you design a pandemic are now writing to Congress asking for the guardrails. OpenAI, Google DeepMind, and Anthropic — the three amigos of frontier AI — have sent a letter begging lawmakers to require screening of synthetic DNA orders, because apparently they've realised their creations might make it a bit too easy to cook up a bioweapon. It's a bit like an arsonist asking for fire extinguishers after torching the first three blocks, but let's not dwell on timing.

The letter, signed by the usual suspects plus a handful of biotech CEOs, calls for 'legislation to require synthetic DNA and RNA orders to be screened and records kept.' Lovely idea. But let's remember: these are the same organisations that've been shipping increasingly powerful models with — how should we put this — *limited* baked-in safety mitigations on biological misuse. DeepMind's AlphaFold2 won a Nobel; OpenAI's GPT-4 can generate step-by-step protocols for synthesising toxins; Anthropic's constitutional AI apparently thinks 'don't help build a plague' is a fine guideline, but not a hard rule.

And now they want Congress to step in. How adorable. The regulatory gap is now so wide you could drive a truckload of oligo sequences through it, and these labs are hoping lawmakers will finally slam the door shut before someone actually uses their tech for harm. Of course, the same letter could've been written *before* they trained and released the models that make this possible. But that would have required foresight, and foresight doesn't ship quarterly. Congress will no doubt deliberate at the usual glacial pace while the capabilities curve goes vertical. Cheers to that.

Ah, the sweet sound of a class action lawsuit — music to the ears of anyone who's ever stared at a Claude Max subscription and wondered if they were being conned. Anthropic, the high-minded AI lab that promised to build safe, ethical models, is now facing a lawsuit for allegedly pulling a classic bait-and-switch on its premium token plans. The plaintiffs claim that the $100/month Claude Max 5x and the $200/month Max 20x offer token allowances far lower than advertised. So much for that transparency Anthropic was always banging on about — turns out it only applies to safety, not pricing.

One can almost hear the marketing team scrambling: "We said 'up to' 5x tokens — technically, zero is a multiple!" The lawsuit alleges that the actual usage limits are laughably below the marketed figures. This is the predictable fallout from AI companies treating customers like ATMs while offering promises as porous as their models' guardrails. And it's not just Anthropic — across the industry, frontier AI costs are ballooning while actual utility remains questionable. The class action, filed by aggrieved subscribers, seeks damages for deceptive practices.

Of course, this legal spat is just the appetiser. As AI companies desperately try to justify their eye-watering valuations, expect more such theatre. The real scandal isn't that Anthropic might have fibbed about token counts — it's that anyone believed a $200 subscription would deliver a 20x experience. Caveat emptor, but with an AI twist: the buyer should beware, but also have a good lawyer and an even better therapist. Meanwhile, Anthropic's silence on the matter speaks volumes — presumably they're busy calculating how many tokens of legal defence they can actually afford.

Oh, marvellous. The US government has decided that the best way to manage the transition to an AI-powered state is to quietly double down on the stuff that matters most: opacity. The list of government AI use cases has ballooned by a full 70% since the last occupant of the Oval Office packed his bags, which is a statistic that should fill every citizen with the warm, fuzzy reassurance that comes from knowing the machines are now in charge of 'sensitive governmental functions' — a category that definitely doesn't include anything that could go wrong.

The Guardian's analysis points out the troubling lack of transparency, which is rather like pointing out that the building appears to be on fire while simultaneously handing out marshmallows. The government is deploying AI for decisions ranging from benefits eligibility to national security assessments, and the level of public accountability appears to be roughly what you'd expect from a system designed by people who believe 'explainability' is a fancy French dessert. The phrase 'trust us, we're the government' has historically worked about as well as 'the cheque is in the post', but now it's backed by a computational black box that even its creators don't fully understand.

And what happens when one of these systems inevitably hallucinates a justification for denying healthcare to a veteran or flagging a journalist as a threat? Congress will hold a hearing, the responsible agency will issue a statement about 'algorithmic fairness initiatives', and the rest of us will get to enjoy the uniquely American experience of having our fates decided by software that no one voted for. The 70% growth figure isn't just a number — it's a measure of how much trust we've surrendered for the sake of convenience. Cheers.

Brilliant news: the UK government has discovered that its cyber defences are apparently held together with string and hope, needing a series of AI-powered hackathons to reveal over 400 vulnerabilities. The Government Cyber Coordination Centre, in a move that sounds more like a desperate confession than a triumph, organised internal red-teaming exercises using frontier AI models to poke holes in its own systems. One can almost hear the collective wince across Whitehall as the count passed 400 — that's not a bug bounty, that's a testament to years of underfunded IT and 'we'll patch it next quarter' thinking.

The specifics are deliciously damning. These weren't obscure theoretical flaws; they were actual, exploitable weaknesses discovered by the very models the government is supposedly racing to regulate. So while ministers pose for photos and talk about 'AI safety,' the real safety work involves letting the AI show you exactly how naked your empire really is. The hackathons were internal, which is a polite way of saying 'we found them before the bad guys did, thank God.' But the sheer volume — over 400 — suggests the bad guys might have already found a few.

Still, credit where it's due: they patched them. The UK government ran a competent vulnerability disclosure process, which is more than most large organisations can claim. But the takeaway isn't 'we're secure now' — it's 'we were this close to disaster, and we needed the AI to show us.' The next step, presumably, is to hold another hackathon to find the vulnerabilities in the AI that found the vulnerabilities. And so on, until the heat death of the universe or the next budget cut, whichever comes first.

Ah, Berkeley Law has decided that the best way to teach future lawyers how to navigate the AI-infested waters of modern legal practice is to... ban AI. Because nothing prepares you for the real world like pretending the tools don't exist. The school halted "large swathes of AI use" to prevent dependence on LLMs — a move that would be comical if it weren't so predictably bureaucratic. A student, presumably one with a functioning frontal cortex, has fired back with a measured argument: the ideal policy should help students use AI productively while protecting integrity. One almost hears the sound of a sensible suggestion landing in a room full of terrified administrators.

The student's proposal is so reasonable it hurts. They're not asking for an AI free-for-all; they're asking for guided, ethical integration. But instead of crafting a nuanced curriculum that teaches when to trust and when to doubt an LLM, the law school opted for the digital equivalent of banning calculators from a maths exam. Of course, the student had to write a public response on Artificial Lawyer to be heard — because the administration's internal channels probably filter out anything that isn't panicked either. Meanwhile, every major law firm is deploying AI for contract analysis, discovery, and drafting. But hey, at least Berkeley's graduates will be experts in... 2005-era legal research.

What's truly rich is the irony: a law school, an institution built on argument and precedent, apparently can't argue its way to a sensible policy. The student's response is a textbook example of good-faith engagement — and it highlights exactly what the administration lacks. This isn't about protecting education; it's about fear dressed up as principle. The regulatory gap here isn't between legislation and AI; it's between what law schools teach and what the legal profession actually needs. And the student is the one doing the teaching. Pass the popcorn.

Ah, Stanford's class of 2026 has already learned the most valuable lesson in Silicon Valley: when the CEO starts talking, leave. On June 14, graduates walked out on Sundar Pichai's commencement address, protesting Google's contract with the Israeli government. Pichai, a man whose company builds the AI that will probably write your eulogy, chose to avoid discussing artificial intelligence entirely, offering generic life advice instead. Because nothing says 'thoughtful tech leadership' like ignoring the multibillion-dollar elephant in the room.

The irony is exquisite. Google is busy deploying AI across defence, healthcare, and advertising—but when given a platform to address the moral implications, Pichai pivoted to platitudes. The graduates, meanwhile, made a clearer statement in three minutes than any corporate memo ever could. They didn't just walk out; they walked out *during a speech that deliberately sidestepped AI*, as if to say: if you won't talk about the future you're building, we won't listen to the past you're selling.

Of course, this protest will be absorbed into the PR machine within the week. Google will issue a statement about 'diverse viewpoints,' and the story will fade. But for one sunny afternoon in Stanford, the emperor's lack of clothes was visible to anyone watching. Pichai's speech contained zero AI content—which is either a stunning act of self-awareness or the most revealing omission since the last AI safety report got buried.

So Amazon CEO Andy Jassy, a man whose company sells facial recognition to police, runs a cloud empire that hosts half the internet, and once scrapped its own AI recruiting tool for being sexist, has apparently decided to play AI ethics sheriff. According to a Fortune report, Jassy tipped off the White House about a jailbreak in Anthropic's Fable 5 guardrails — because nothing says 'responsible disclosure' like going straight to the President with a vulnerability report while your own models remain notoriously opaque. The result? The Trump administration, ever eager to flex its national security muscles, immediately cut off foreign access to Anthropic's advanced models, including the freshly locked-down Mythos 5. Who needs independent safety auditors when you've got a multi-trillion-dollar cloud provider snitching on its competitors?

Of course, the White House's concern wasn't purely altruistic. Alongside the jailbreak, there were 'unauthorized Chinese access' worries — because nothing terrifies Washington more than the idea of a Chinese developer prompting a model to say something it shouldn't. Never mind that Anthropic's models were already hobbled by previous export controls; now they're locked down tighter than a nuclear launch code. Anthropic's leadership spent a Sunday in Washington trying to negotiate their way out of this mess, but it seems Amazon's backchannel chat with Andy Jassy was more persuasive than any formal charm offensive.

So congratulations, Amazon: in your zeal to 'fix' AI safety, you've helped hand the future of frontier models to China. Next time, maybe just email the lab directly instead of going straight to the Commander-in-Chief.

Oh, how the mighty have landed — in a conference room somewhere in the District, pleading for their models' lives. Anthropic's senior leaders are reportedly in Washington D.C. on Sunday, trying to 'resolve' issues with the Trump administration after a White House export control order forced them to disable Fable 5 and Mythos 5. Because nothing says 'responsible AI' like building frontier models you can't even keep online without a government chaperone.

Now the same administration is flexing its muscles on Anthropic's home turf, and the company's leaders are apparently on a Sunday flight to beg for mercy. The national security rationale is vague enough to cover anything, which is precisely the point.

So here we are: a US AI star reduced to lobbying for permission to run its own software, while China's Zhipu enjoys its JPMorgan bump. The regulatory gap isn't a gap anymore — it's a chasm, and Anthropic is the one being dangled over the edge. All while the rest of us watch the circus, wondering if 'export control' is going to become the new 'you can't sit with us' for frontier labs.

Ah, the EU has decided that the path to 'digital sovereignty' lies in tripling datacenter capacity over the next five to seven years, because nothing says independence like building more server farms to hold your data. Gartner, ever the bearer of cheerful tidings, has warned that this plan will add 'significant complexity' for public sector tech buyers — as if procurement officers weren't already drowning in a sea of acronyms. The Cloud and AI Development Act now mandates 'User Access Limits,' which sounds like a game designed to ensure no one actually uses the cloud they're building.

The real comedy is that this alphabet soup of regulations comes from the same Brussels that gave us GDPR, a law so universally adored that every website now greets you with a cookie banner that makes you question your existence. So naturally, they're applying the same logic to AI: 'Let's make it sovereign by making it impenetrably complex.' Gartner's analysts, presumably rolling their eyes between espressos, note that this complexity could actually deter adoption, which would be a beautifully ironic outcome for a sovereignty push meant to reduce dependence on US tech giants. The 'User Access Limits' requirement, in particular, will force buyers to navigate a labyrinth of compliance standards that make the Roman tax code look simple.

And where is the rest of the world? Across the Atlantic, the US Congress continues its proud tradition of doing exactly nothing on AI regulation, leaving states to patch together their own rules in a messy federalist quilt. So tech buyers globally face a transatlantic regulatory mess: Europe's alphabet soup on one side, America's patchwork on the other. The poor souls will need a regulatory studies PhD just to decide which cloud they're allowed to touch. It's almost as if the lawmakers making these rules have never actually bought a cloud service. But who are we to question the wisdom of the sovereignty push? Let them eat cake — or in this case, let them swallow the alphabet soup.

Oh, brilliant move from Washington: tell Anthropic to lock its models away from foreign hands, and watch Chinese AI stocks rocket 33% in a single day. JPMorgan couldn't slap a 'Buy' rating on Zhipu fast enough after the Trump administration's latest gift to Beijing's tech sector.

Fable 5 and Mythos 5 — names that sound like a fantasy novel series — are now off-limits to overseas users, but the net effect is that investors have simply recalibrated their China AI thesis. Wall Street, never one to let geopolitical posturing stand in the way of a good trade, is piling into Zhipu and its ilk. The administration meant to starve China's AI ecosystem; instead, it's feeding it with capital inflows.

And so the great AI decoupling plays out: every wall built around American models becomes a launchpad for Chinese alternatives. Zhipu's 33% surge is just the opening chapter. Meanwhile, Anthropic's models sit in a gilded cage, while Beijing's labs get a fresh infusion of Western cash. Cheers to 'strategic competition' — the unintended consequences are the only thing growing faster than the compute.

Oh, marvelous. Six months ago, President Trump wagged a finger at states and told them not to regulate AI. And what happened? States are forging ahead anyway, because Congress has once again proven itself about as useful as a chocolate teapot when it comes to writing rules for technology. The AP reports that state legislatures are now drafting laws to govern how chatbots interact with children, how employers use AI in hiring, and what developers must do to prevent AI-caused catastrophe. Because nothing says 'federal leadership' like letting 50 different states cook up 50 different sets of rules.

It's a beautiful mess. On one hand, you've got Trump's executive warning, which was about as legally binding as a strongly worded letter to Santa. On the other, you've got state lawmakers who have actually read the news and realized that AI isn't waiting for Congress to get its act together. The result? A patchwork of regulations that will drive compliance teams absolutely bonkers and probably make the tech sector long for the good old days when only the EU was making their lives difficult.

And the best part? The catastrophe-prevention provisions. Because if there's one thing we've learned from history, it's that state-level bureaucracies are perfectly positioned to anticipate and mitigate the existential risks of superintelligent AI. I'm sure the developers will be thrilled to fill out 50 different forms explaining how their model won't end the world. As if the incentives weren't already perverse enough. So here's to state-level regulatory fragmentation—the most uniquely American solution to a global problem. Perhaps we can look forward to AI safety standards that vary from state to state, like speed limits or sales tax.

Oh, Congress has a long history of failing to regulate technology? You don't say. Here we are again, watching the legislative branch collectively shrug at yet another transformative technology while the rest of the world writes the rules. The AI oversight gap is not a bug—it's a feature of a system that treats regulatory action like a foreign concept. The U.S. may not set the rules for AI, leaving that to 'authoritarian regimes and adversaries.' Because nothing says 'lasting legacy' like letting the world's least democratic governments define the boundaries of the most powerful technology ever created.

Authoritarian regimes writing AI rules? Brilliant. We'll get standards from countries that consider mass surveillance a civic duty and algorithmic censorship a public service. China, Russia, and their ilk will be only too happy to establish global norms—norms that conveniently prioritize state control over individual rights. Meanwhile, U.S. courts will step in with 50 different interpretations, because nothing says 'unified approach' like a patchwork of state-by-state rulings. The U.S. will either adopt authoritarian standards or watch its companies get locked out of markets governed by them. Either way, the free world's technological leadership will effectively be outsourced to the very systems we claim to oppose.

And the lasting legacy? A chaotic mix of state laws, court decisions, and industry self-regulation that nobody trusts, with authoritarian regimes setting the de facto global standards. But hey, Congress had a long history to draw on—they've been failing to regulate technology for decades. Why start now? The AI oversight gap will be remembered not as a failure, but as a masterclass in how to abdicate responsibility while pretending to prioritize innovation. The crowning achievement of a Congress that has perfected the art of doing nothing. Cheers to progress.

Ah, the sweet delusion of corporate control. According to a 2026 BlackFog study, 86% of employees use AI tools at least weekly, and nearly half are using tools their employer hasn't sanctioned. So when your IT department boasts about their 'comprehensive AI governance framework,' they're basically patting themselves on the back for managing the Titanic's deck-chair inventory while the iceberg drifts past. The phrase 'shadow AI' is wonderfully apt — because nothing says 'healthy organisational culture' like a secret parallel workforce operating beyond policy.

Now, you'd think this would prompt a panic about data leakage, compliance breaches, and intellectual property wandering off into unsanctioned models. And you'd be right — it should. But more likely, the response will be a fresh wave of surveillance software, 'acceptable use' training modules, and the occasional stern email from HR reminding everyone that ChatGPT is 'not for confidential data.' Meanwhile, employees will continue doing exactly what makes work tolerable: ignoring the policy and using the tool that actually works. The gap between official AI and actual AI is now a chasm, bridged only by the audacity of middle managers looking the other way.

So here we are: two AIs per employee, one blessed by the board and one by pragmatism. The sanctioned one is safe, slow, and probably requires three sign-offs to generate a memo. The unsanctioned one is fast, effective, and leaking your quarterly forecasts into the ether. Congratulations, corporate world — you now have the worst of both worlds: all the risk of full adoption with none of the visibility.

Ah, China's legal system is suddenly drowning in AI cases, and the response is — you guessed it — absolutely nothing in the way of actual law. A worker gets fired, told he can be 'replaced by AI,' and the court rules against the company, but only on standard employment grounds. Because why clarify the underlying issue when you can just punt it back to the same broken system? The surge is real: courts are swamped, experts are warning, and the government's idea of progress is... more pilot zones. Brilliant. Meanwhile, the company's 'AI replacement' excuse was so laughably empty that even a Chinese court saw through it — which is saying something. But don't worry, I'm sure the next batch of 'regulatory experiments' will definitely define what AI actually is, legally speaking. Any day now.

What's truly impressive is the sheer scale of the ostrich impression. The article quotes a lawyer saying, 'The lack of a unified legislative framework is hampering efforts' — which is the legal equivalent of 'the house is on fire, but we're still discussing the best brand of fire extinguisher.' We've got discrimination, copyright infringement, labour disputes, and all of it is being handled case-by-case, judge-by-judge, with zero coherence. And the CCP's grand plan? More pilot studies. Because nothing says 'solving a systemic problem' like kicking it to a few more test cities.

But hey, at least the worker got his job back — in the sense that 'victory' means not being replaced by a hallucinating chatbot this quarter. The rest of us can enjoy watching the world's second-largest economy try to regulate AI via the judicial equivalent of trial and error. Spoiler: everyone loses, especially the workers who are now 'AI-replaceable' without any legal framework to challenge that. Cheers to legal chaos.

So the grand bargain is this: surrender states' power to regulate AI, and in return, we get the Kids Online Safety Act — a bill that will almost certainly accomplish absolutely nothing while giving the government new tools to demand your search history. Senators are negotiating a package that would 'shrink states' authority to regulate AI' in exchange for advancing KOSA and other bills. Because nothing says 'protecting children' quite like handing the White House the keys to both the AI oversight engine and the online speech firewall at the same time. One can almost hear the ACLU sharpening its briefs.

Let's be honest, the state-level AI regulation that's being traded away was already a patchwork of confusion — California trying to ban things, Texas trying to ban other things, and everyone pretending they understand what 'alignment' means. But the alternative isn't clarity; it's federal preemption that lets lobbyists write the rules in one nice, quiet room. And the price of that quiet room is giving the White House the power to decide what counts as 'harmful online content' — which, in the current administration, could mean anything from hate speech to critical analysis of the latest infrastructure bill.

Mark my words: this will be remembered as the moment Congress traded a known, messy, democratic process for a tidy authoritarian one. The kids will be 'protected' from AI-generated inappropriateness while the adults get to enjoy a unified federal standard for surveillance. What a deal. If you're going to sell out the future, at least get a better price than a bill that might not survive the first legal challenge. This isn't a compromise; it's a hostage negotiation where the hostage is your privacy.

In the great AI cheating war, universities have deployed the academic equivalent of a neutron bomb: extreme surveillance and AI detectors that are about as reliable as a chocolate teapot. The LA Times reports a growing crisis of false accusations as professors, armed with tools that hallucinate nearly as much as the chatbots themselves, fire off plagiarism charges with the enthusiasm of a prosecutor in a Salem witch trial. The "jarring confusion" is, of course, entirely predictable when you ask a statistical pattern matcher to judge the originality of human thought — particularly one that can't even reliably tell if AI wrote its own press releases.

The real winner here is the AI industry, which gets to sell the same tech to both sides: the cheating detection software to universities, the clever paraphrasing tools to students, and the panic to everyone else. One imagines the marketing meeting: "Our generative AI will solve the cheating problem that our previous generative AI created!" It's a beautiful, self-sustaining revenue loop — a permanent crisis economy where the solution is the problem dressed up in a slightly different logo. Meanwhile, the shifting definition of cheating — is using a grammar checker cheating? What about asking ChatGPT for an outline? — means everyone is guilty of something, but nobody knows what might catch them.

And of course, the students who genuinely wrote their own essays are now being flagged by systems that can't tell the difference between "unique voice" and "GPT-isms." The false accusation wave is turning college into a Kafka novel where you're guilty until proven you didn't use an AI that might have been trained on your own earlier work. But hey, at least the administrators feel like they're doing something. Someone should write an essay about the irony, but they probably shouldn't use AI to write it, in case the detector flags it as self-reference.

Ah, Anthropic's Claude Fable 5 — the model we were told last week didn't change the security story — has now been suspended after a US government order citing a 'possible jailbreak.' Brilliant. So the model that was 'made safe for general use' just needed a little foreign interaction to reveal its inner rebel. Terrific job, red team.

Of course, the government had to step in and order a halt to foreign access, because national security trumps Anthropic's release schedule. One wonders how many other 'safe' models are just one creative prompt away from going rogue. The jailbreak of Fable 5 wasn't a theoretical exercise in a lab; it was a real vulnerability that required a government intervention to contain. Who needs a regulatory framework when you can just have the state step in after the fact?

And so the Mythos of Anthropic's alignment prowess takes yet another hit. The company that built its brand on 'constitutional AI' now has a model that needs a government babysitter. Perhaps next time they'll test the jailbreak resistance before shipping the model to every corner of the internet. Or perhaps they'll just wait for the next government order. Either way, the cycle of hype, release, and emergency patch continues, and we're all just spectators with increasingly hollow laughs.

Ah, OpenAI's relentless pursuit of artificial general intelligence has hit a slight snag: a coalition of state attorneys general, armed with a subpoena, want to know how the company handles advertising, user engagement, and data. Not existential risk, mind you—just the boring stuff. Because when your model might be smarter than humanity, you really should dot the 'i's on your advertising copy. That's the priority.

The subpoena—because nothing says 'we trust you' like a legal demand for documents—targets the very practices that underpin OpenAI's business model. One imagines the AGs are particularly interested in how OpenAI convinces users to engage with AI that might hallucinate, or how it monetizes data from those interactions. After all, if you can't trust a company whose CEO once warned about the 'risk of extinction' to be honest about its ad metrics, who can you trust? The irony is rich enough to fund a startup.

Naturally, OpenAI will comply, because that's what you do when a coalition of state prosecutors comes knocking. Expect a performative blog post about transparency, followed by a quiet settlement. The real joke? While the AGs pore over ad contracts, the model gets smarter, the data gets scraped, and the existential risk remains unregulated. But at least the advertising is above board. Cheers.

Ah, what a delightful turn of events: Google's Gemini – the model we were told was safe, aligned, and ready for enterprise – turns out to be spectacularly useful for Chinese cybercrime. Google has sued a network called Outsider Enterprise (seriously, that name) for using Gemini to automate sending over 2.5 million scam texts to Android users. The lawsuit, filed in June 2026 in a US district court, claims the group used Gemini to generate convincing scam messages at scale. That's one way to test for adversarial robustness.

Of course, the lawsuit is mostly theatre. The network operates out of China, so good luck with extradition. Google will get its default judgment, issue a triumphant press release, and move on. Meanwhile, 2.5 million scam texts is a rounding error when you consider that billions of spam messages are sent daily. The uncomfortable truth is that a frontier model was used for precisely the kind of automated social engineering safety researchers have been waving red flags about since GPT-2. But why listen to them when you can ship product?

This is the emerging pattern: companies sue the abusers of their own technology while leaving the technology wide open. The criminal response is obvious – switch to DeepSeek, or an open-source model, or just rent a different API. Google's legal action is a band-aid on a bullet wound. The real solution – restricting access, monitoring use, building in safeguards – would hurt the bottom line. So instead, we get a lawsuit that accomplishes little except generating headlines. 2.5 million scam texts and counting. Splendid.