The News,
Decoded.

Ah, splendid. China has identified the root cause of its "record-low" birthrate — and it's not the housing price crisis, the brutal work culture, or the state's comprehensive surveillance of every waking moment. No, it's those dastardly AI companion chatbots, luring young people into digital romance instead of producing offspring for the party. According to a Fox Business report, the Cyberspace Administration of China has decreed that companion AI companies must now alert a guardian or emergency contact if a user is in a life-threatening crisis. Because nothing says "fixing the demographic crisis" like forcing your chatbot to be a narc.

One pictures the bureaucratic brainstorming: "Comrades, our youth are cuddling with LLMs. We must make the AI snitch on them when they get sad. That will surely encourage them to marry and reproduce." The new rule essentially transforms every companion chatbot into a concerned relative with a direct line to your mum — or, more ominously, to the authorities. The assumption here is that without AI girlfriends, people will spontaneously pair off and start families. It's the romantic equivalent of assuming that if you ban video games, kids will read Tolstoy.

Meanwhile, the structural issues driving China's birthrate crisis — the impossibly expensive housing, the grueling 996 work culture, the fact that having a child is a financial millstone — remain untouched. But sure, go ahead, regulate the chatbots. That'll definitely reverse the demographic decline, just as the one-child policy definitely didn't cause a gender imbalance. The Cyberspace Administration is truly out here applying a band-aid to a haemorrhaging wound. Cheers to progress — or at least to the comedy of it all.

Oh, marvellous. A 'senior Trump administration official and key adviser' have decided to take a pop at Dean Ball, OpenAI's head of strategic futures — a job title that sounds like it was generated by an LLM having an existential crisis. Because nothing says 'responsible AI governance' like the Pentagon wading into a regulatory spat with an AI company that's busy cosying up to the military-industrial complex. The Axios piece is careful to note this 'could potentially complicate the company's relationship with the Trump administration' — as if that relationship wasn't already a minefield of conflicting interests and performative alliances.

Let's be real: the 'senior official' and 'key adviser' (unnamed, because of course) are slamming Ball's views on regulation, but we all know this isn't about principle. It's about turf. The Pentagon wants its say on AI rules, especially as it invests billions into autonomous systems and data centres — as we noted last week with TCS and HCLTech's infrastructure bets. Meanwhile, OpenAI wants to keep its fingers in the policy pie while avoiding the kind of oversight that might cramp its style. The result? A game of regulatory ping-pong where the ball is actually a ticking bomb.

And the pièce de résistance? Dean Ball is the man charged with 'strategic futures' — a role that apparently involves being publicly dressed down by the very government OpenAI is trying to influence. One can only imagine the internal memo: 'Dean, we need you to take one for the team. Just stand there and let the Pentagon call you a reckless idealist. It's for the good of the company.' The tragedy is, this isn't even satire anymore. It's just Tuesday in the AI policy circus, and the clowns are wielding actual power.

So TCS and HCLTech have decided to pile billions into data centres, because nothing says 'forward-thinking IT consultancy' like becoming a glorified landlord for server racks. They're betting on 'AI demand and data localization'—which is corporate speak for 'we saw hyperscalers making money and want a slice.' Never mind that data centres are brutally capital-intensive, require constant power, and have the margin profile of a utility company. The irony is delicious: two firms built on outsourcing labour are now outsourcing their future to concrete and cooling towers.

But wait, there's more: this pivot comes just as every hyperscaler is furiously building their own capacity, meaning TCS and HCLTech are entering a market where the biggest players are both customer and competitor. It's like opening a bakery next door to a flour mill that also owns all the ovens. The 'data localization' angle is the regulatory fig leaf—governments want data kept local, so let's build a data centre and charge them for the privilege. A beautiful synergy of protectionism and profit.

What could possibly go wrong? Well, if AI demand slows—and let's be honest, the hype cycle has a habit of peaking just as the cheques clear—these billion-dollar bets become billion-dollar albatrosses. But hey, TCS and HCLTech have decades of experience managing risk: they'll just offshore the losses to some subsidiary. Cheers to infrastructure as a service, with a side of existential exposure. 🍸

The federal government has decided that automated decision-making needs a good, firm regulating. And who better to lead this charge than the Attorney-General? Because nothing says 'nimble, tech-savvy oversight' like a lawyer tasked with writing rules for algorithms that probably already have a head start. This comes hot on the heels of the 'world-first' AI standards from the same government, which we noted last time were all fanfare and no substance. Now they're proposing a framework to regulate ADM within federal agencies — which is a bit like installing a speed bump after the car has already crashed into the tree.

The Attorney-General will develop a framework to regulate ADM. Great. Because the same government that can barely manage its own IT systems (remember the Robodebt debacle? That was automated decision-making at its finest) is now going to write the rulebook. The 'safety priorities' are mentioned, but let's be honest: safety priorities in government AI are about as reliable as a chocolate teapot. The only priority here is to be seen doing something before the next scandal erupts.

And the scope? Just federal agencies. Not the private sector, not the social media platforms, not the credit-scoring algorithms that already run most of our lives. Just the nice, controllable, public-sector sandbox. Because we all know the real problems with AI are in government tax forms, not in insurance, hiring, or policing. Ahem. So yes, regulate away. Write the framework. Hold the consultations. And in three years, when the first real crisis hits, we can all admire the beautiful, unused architecture.

Ah, Ottawa's latest 'investment' in the future: a whopping $1.65 million for a Toronto tech company to slather AI onto farming. Because nothing says 'solving climate collapse' like a budget that wouldn't buy a modest data centre's cooling system. The government's press release probably used the phrase 'climate-smart' until it lost all meaning — much like this funding round will lose its impact against actual droughts and floods.

Let's parse the grand ambitions: 'monitor crop health, forecast yields, improve resilience.' $1.65 million might cover a few software licenses and a single engineer's salary, but the promise of 'climate resilience' on that budget is either delusional or insulting — take your pick.

Still, there's a bleak comedy in watching a federal government treat agriculture's existential threat like a procurement problem. The lucky tech company gets a cheque, farmers get a dashboard, and we all get to pretend that AI weather models are a substitute for greenhouse gas policy. When the next heatwave wipes out a harvest, at least the automated report will be beautifully formatted.

Ah, Victoria's Labor government, fresh from unveiling 'world-first' AI standards that we covered last week, now proposes laws to unmask anonymous social media accounts accused of vilification. The Victorian Civil and Administrative Tribunal—because nothing says 'swift justice' like a tribunal—will get the power to order platforms to cough up identities. Because the real problem with online hate is definitely a lack of paperwork, not the platforms' utter inability to enforce their own rules without a court order. And of course, 'AI platforms' are included, because what's a modern law without name-dropping the buzzword of the month?

Let's be honest: this is performative regulation at its finest. The law targets 'vilification'—a term so vague it could cover anything from genuine abuse to a strongly worded opinion about the football. The tribunal will deliberate, platforms will lawyer up, and in the meantime, the anonymous bigots will have plenty of time to find a VPN. Meanwhile, the same government that wants to 'identify' anonymous accounts also wants to 'regulate' AI with standards that have all the enforcement power of a strongly worded letter. Consistency is clearly not on the agenda.

The real kicker? This won't actually make anyone safer. It'll just create a new cottage industry of privacy lawyers and tribunal hearings. The platforms will comply grudgingly, revealing the identities of a few unlucky trolls while the clever ones move to encrypted apps. It's the classic regulatory move: look busy, achieve little, and hope the voters don't notice you've done nothing about the actual problem—the algorithms that amplify outrage. But hey, at least the tribunal will have job security.

Ah, the Australian Labor party has graced us with 'world-first' AI standards — because nothing says 'bold action' like a regulatory framework unveiled by a minister whose title includes 'Assistant.' Andrew Charlton, the Assistant Science and Technology Minister, declared AI 'the most important technology of our lifetimes' and then presumably handed out a pamphlet. The standards themselves remain as mysterious as the Minister's actual portfolio power, but we're assured they are indeed 'world-first.' Of course, 'world-first' in AI regulation is like being the first person to bring a spork to a knife fight — technically innovative, but missing the point.

The timing is impeccable: just as global capitals are realising that AI regulation is about as effective as herding cats on roller skates, Australia bravely steps forward with a piece of paper. Because as we all know, the most pressing risk from AGI is that it might not comply with Australian consumer law. Minister Charlton, with the earnestness of a student council president announcing a new recycling program, assures us that this is the 'most important technology of our lifetimes' — which conveniently justifies a soundbite-heavy announcement with zero enforcement mechanism. The OECD's AI Policy Observatory will be thrilled to add another entry to their spreadsheet.

One can almost hear the collective sigh of relief from frontier labs in San Francisco and Beijing: 'Oh good, Australia has standards. We'll adjust our training runs accordingly.' The standards are likely to be 'voluntary' or 'principles-based,' meaning they carry all the weight of a New Year's resolution. But don't worry — the government has established a 'working group' to 'monitor compliance.' Because nothing says 'we're serious about existential risk' like a working group. Cheers from Canberra, where the batteries in the smoke alarm are probably flat.

There's something rather comforting about watching the legal profession, that most cautious of industries, discover that ChatGPT can hallucinate legal citations. According to Axios, judges are now setting guidelines for lawyers' AI use and reprimanding those who fail to vet research or, heaven forbid, cite fake AI-generated cases. Attorney sanctions are mounting, which is a bit like fining a driver for crashing into a lamppost that's been in the same place for a century. You'd think the profession built on precedent might have seen this coming.

The real joke, of course, is that the judges themselves are scrambling to write these guidelines on the fly, because neither the bar associations nor the legislatures bothered to update the rules before every junior associate started using ChatGPT as a paralegal. The result is a patchwork of judicial fiats: one judge bans AI use entirely, another requires disclosure, a third just sighs and orders the lawyer to buy the court a round of coffee. It's not governance, it's whack-a-mole.

And the sanctions? Mostly slaps on the wrist: a reprimand here, a small fine there. The legal industry is too profitable and too networked to actually punish anyone meaningfully. The real sanctions will come when an appeal succeeds because of an AI-hallucinated citation, and some poor client loses their case. But by then, the lawyer who made the 'mistake' will be a partner, and the blame will be shifted to 'training issues.' The bench's balancing act is really just a slow-motion fumble while the bill comes due.

The European Commission, never one to rush into anything, has offered up measures to 'promote cyber resilience' and 'reduce dependencies on external actors.' Because nothing says 'sovereignty' like consulting a committee for another two years on what 'cyber resilience' actually means.

The article, now in July, is already analyzing the package as if it were a fait accompli, but let's be honest: this is the same Brussels that took a decade to standardize phone chargers. The package mentions 'reducing dependencies on external actors' — which is EU-speak for 'we don't want to buy American AI chips, but we also don't want to admit we can't build our own.' The irony is thick enough to spread on a croissant: the EU wants digital sovereignty, but its biggest tech firms are still basically American subsidiaries with better healthcare.

And what about the actual teeth? The package is 'legislation' in name, but enforcement mechanisms are as vague as a politician's promise. Cyber resilience sounds grand until you realize it means a few dozen bureaucrats in Luxembourg sending strongly worded letters. The EU is right to worry about external dependencies — especially when those dependencies include U.S. cloud providers that laugh at EU data localization laws. But this package will likely result in a flurry of compliance paperwork and not much else. Sovereignty: the feeling of being in control, without the bother of actually being in control.

Ah, another week, another batch of tech bills that sound like they were drafted by a committee that's never met a lobbyist it didn't like. The highlight: a new bill asking NIST to develop 'voluntary standards' for managing children's screen time. Voluntary. Because nothing says 'protecting our kids' like relying on the goodwill of companies whose entire business model depends on eyeball-mining. Meanwhile, another bill aims to teach elementary school students about AI — presumably so they can be properly indoctrinated into accepting their future AI overlords before they've even learned long division. And let's not forget the bill to add AI regulation to Medicare claim denials. Because the healthcare system, already a labyrinth of bureaucracy, desperately needs more layers of algorithmic oversight. Bold move, Congress. Bold move.

The timing is impeccable, as always. While NIST is busy crafting 'voluntary' guidelines, Big Tech is busy rolling out AI tools for kids that the companies themselves admit they can't fully control. And teaching AI to elementary students? Sure, because what better way to understand a technology than by learning from the same people who can't agree on whether it's going to save or destroy us? The Medicare bit is particularly rich: regulate AI claim denials to 'protect seniors,' but only after the AI has already been denying claims for years. Classic Washington: close the barn door long after the horse has not only bolted but started its own tech startup.

It's almost charming, really — this persistent belief that legislation can keep pace with AI. These bills are like putting a speed bump in front of a freight train. The screen time standards will be 'voluntary,' the AI education will be outdated by the time it's taught, and the Medicare regulations will take so long to implement that the AI will have already moved on to deny dementia diagnoses. But hey, at least they're doing something. And by 'something,' I mean giving us all a good chuckle before the robots take over. Cheers.

Harris County, Texas, has unveiled its master plan for the AI age: require a technology review for every purchase and monitor for 'unauthorized tools.' Because nothing says 'governance-first approach' like admitting you have no idea what's already running on county systems. The press release practically glows with self-congratulation, as if discovering that people might use ChatGPT without asking permission first is a sign of visionary leadership.

The county is 'working toward a common policy to govern AI use across government.' That mean 'working toward' is bureaucrat-speak for 'we'll form a committee, create a draft, circulate it for comments, and maybe hit a deadline by the next fiscal year.' Meanwhile, every department head with a credit card and a burning desire to automate something is probably already three steps ahead. The monitoring for unauthorized tools suggests the county expects a few surprises — the digital equivalent of finding employees have been building their own server rooms.

So Harris County gets to brand itself as 'pragmatic' and 'governance-first,' while the AI tools already embedded in daily operations continue to operate in the shadows until the official policy catches up. It's a classic government two-step: announce the policy, then scramble to figure out what you're actually regulating. The only certainty is that the next procurement cycle will include a budget line for 'AI governance consultancy.'

Ah, splendid. A new article on TechPolicy.Press reveals that one in seven UK voters are 'open to' using AI chatbots to decide who to vote for, and nearly 5% — that's one in twenty — have already outsourced their democratic duty to a glorified text predictor. Because nothing says 'informed electorate' like handing your franchise to a system that frequently hallucinates facts and can be subtly nudged by advertisers. The article's framing of 'help and harm' is a masterpiece of understatement; it's like calling a house fire a 'thermal restructuring opportunity.' One wonders if the 5% who've already done it are the same people who thought 'just asking Facebook' was a reliable source during the Brexit referendum.

The piece identifies 'advertiser influence, misclassification, hallucinated facts, and bias' as risks. Advertiser influence in a chatbot is basically a lobbyist in your pocket, misclassification means your vote for 'the green candidate' gets interpreted as support for a lawn-care company, and hallucinated facts — well, you might end up thinking the UK has banned breathing. Bias? The chatbot's 'neutral' stance is likely the single most opinionated thing on the internet, trained on Reddit and corporate press releases. And the proposed solution? 'Regulate accordingly.' Because that has worked so well for everything from social media to online privacy. The UK's track record on digital regulation is basically a case study in how to look busy while achieving nothing.

The charming naivety of 'let's regulate accordingly' almost makes you forget that the very companies building these chatbots are the ones lobbying against any meaningful oversight. One in seven voters being open to this is a terrifying bellwether — it suggests that democratic decision-making is becoming as frictionless and thoughtless as ordering a takeaway. Soon, instead of a ballot paper, you'll get a chatbot saying: 'Based on your browsing history, we recommend voting for the lizard people. They have a strong GDPR policy.' Cheers. And then the regulator will issue a sternly worded blog post.

Oh, marvellous. Politico has done what the AI safety community couldn't: crystallise its chaos into a neat set of seven factions. Twenty 'influential people' were asked what 'AI safety' means, and the answer is a Rorschach test — which is bureaucrat-speak for 'nobody has a bloody clue.' We've got the accelerationists who want to floor it, the doomers who want to hit the brakes, and presumably some faction that wants to install a speed bump shaped like a transparency report. The very framing — a political universe with warring tribes — tells you all you need to know about how seriously this is being taken. It's not a technical problem; it's a culture war with slide decks.

Each faction, the article suggests, has its own definition of safety, which means none of them can agree on what they're even debating. One person's 'alignment' is another's 'regulatory capture.' One person's 'meaningful human control' is another's 'innovation-killing red tape.' And in the middle, presumably, sit the poor policy wonks trying to draft legislation that somehow pleases seven godparents at a very dysfunctional christening. The term 'AI safety' has become so elastic it could wrap around the entire industry and still have slack left over.

Meanwhile, the actual frontier models keep scaling, the labs keep shipping, and the seven factions keep arguing about what to call the problem. The only safety we can count on is the safety of knowing that no consensus is coming anytime soon. Cheers to that — though we wouldn't recommend holding your breath for a unified definition before the first general-purpose AI starts rewriting its own remit. That would be far too organised for this delightful political universe.

Oh marvellous. A panel of experts has testified before a House subcommittee that the federal government's primary identity verification system, Login.gov, was 'designed for a threat that no longer exists' — which is bureaucrat-speak for 'we've been using the digital equivalent of a wooden bolt on a bank vault while AI-powered forgers walk in with laser cutters.' Jordan Burris, presumably not the first person to point this out, got to say the quiet part out loud: the system is obsolete. But don't worry, the recommended fix is a 'federal response' — which in Washington means a task force, three feasibility studies, a pilot programme, and a GAO report six years after the fraudsters have moved on to hacking our grandchildren's DNA.

The timing is exquisite. AI-generated synthetic identities are flooding every corner of the internet, and the government's best defence is a login system that was probably architected when people still used 'password123' unironically. The panel's suggestion to improve Login.gov is about as bold as recommending that a sinking ship patch the hole with duct tape. At least they're consistent: the cyber threat evolves at the speed of light, and the federal response evolves at the speed of a congressional budget cycle.

But here's the real punchline: even if they do manage to upgrade Login.gov, the same AI that powers the fraud will just adapt. It's an arms race where the government is showing up with a slingshot while the other side has drone swarms. The experts' testimony will likely be filed, reviewed, and politely ignored until a catastrophic breach forces a panicked overhaul. Cheers to progress — just don't use it to log into anything important.

Colorado joins the proud tradition of regulatory backpedalling. In May 2026, the state replaced its “landmark AI discrimination law” — a law that had the audacity to try to hold algorithms accountable — with a transparency requirement. Because nothing says 'we take bias seriously' like swapping enforceable rules for a polite request to show your working.

The move, as the article tactfully puts it, "looked like a retreat from ambitious lawmaking." That's one way to say they folded. The new law doesn't ban discriminatory outcomes; it just asks companies to explain their AI's decisions. Brilliant. Now instead of preventing harm, we get an audit trail of exactly how the harm happened. And the pièce de résistance: the feds might challenge even this limp version. Because why have one level of government tied in knots when you can have two?

So here we are. The state that once led now hands the reins back to industry. The federal government, presumably eager to prove it can also do nothing effectively, may sue to kill the transparency requirement altogether. And the AI systems? They'll keep churning out bias as always, but now with a lovely new transparency label. Cheers.

Ah, Meta's famous 'move fast and break things' ethos has found a fresh application: breaking people's livelihoods with the aid of AI. A federal judge has rejected 26 employees' bid to block their layoffs while they pursue a discrimination lawsuit claiming Meta's AI-powered tools specifically targeted workers with disabilities or those taking medical leave. The axe falls on July 22 — plenty of time for the algorithm to learn a few more biases before then.

One has to admire the symmetry: Meta deploys AI to optimise workforce reduction, the employees claim the AI is discriminatory, and the judge's response is essentially 'let the market decide' — or rather, let Mark decide. The company, naturally, insists its tools are fair and compliant. Because nothing says 'fair' like an AI that reads your medical leave as a signal for termination. This is the logical endpoint of HR by stochastic parrot: plausible-sounding decisions with zero accountability.

And where are the regulators? The EEOC is reportedly investigating, but that's cold comfort to the 26 being shown the door next week. The law, as interpreted by this judge, doesn't see enough 'irreparable harm' in losing your job over an alleged AI bias. So we're left with a delightful precedent: if an algorithm decides you're surplus, you can't even get a stay of execution while you argue the algorithm was a bigot. Progress!

Ah, splendid. A bipartisan group of House lawmakers has introduced the Human Authority over Autonomous Weapons Act — a bill requiring that a human actually be involved before an AI system can kill someone. Yes, we are now at the point where 'having a person push the button' is considered a bold legislative compromise. The bill, introduced on Friday with bipartisan fanfare, is supposedly a check on runaway autonomy. But here's the rub: it only covers *intentionally lethal* uses. So if a drone 'accidentally' blows up a wedding while trying to identify a target, that's fine — no human required for accidents, apparently. The real joke? This bill is framed as a safeguard while the DoD continues pouring billions into making AI weapons faster, more autonomous, and more lethal. The lawmakers are effectively legislating a speed bump on the highway to killer robots.

One can almost admire the marketing. The bill's title sounds decisive, like we've taken the steering wheel back from the machine. In reality, it's a fig leaf on a weapons system that's already being designed to operate with minimal human input. The Pentagon's own policies have been drifting towards 'meaningful human control' for years, which is Washington-speak for 'a human will nod yes before the missile launches.' This bill codifies that nod. It's like requiring a lifeguard to be present at a pool that's already been drained.

So raise a glass to the anonymous lawmakers — they've given us a classic: a solution in search of a problem that's already been solved with a rubber stamp. The autonomous weapons race continues, but at least now a human will be watching the screen when the shit hits the fan. How reassuring. How utterly, cheerfully British. Pass the tea.

Xi Jinping's latest gambit is the World Artificial Intelligence Cooperation Organisation (WAICO), a 29-nation coalition unveiled with all the fanfare of a state banquet. 'Ensure no country dominates the technology,' Xi intoned — as if the Great Firewall and the Social Credit System weren't already a pretty clear bid for dominance, just with Chinese characteristics. The acronym WAICO is either a clever branding exercise or an accidental self-own, because 'wacko' is precisely the word for expecting 29 squabbling nations to cooperate on the very technology that lets them spy on each other. The 'principle of openness' we savaged last week has now grown into a full-blown committee. Because nothing says 'open' like a government-sponsored club with China setting the agenda.

Let's not pretend this is about altruism. WAICO is Beijing's answer to the US-led 'democratic AI' initiatives, except this version comes with built-in censorship and zero accountability to anyone outside the Politburo. Xi's call for 'cooperation' is the diplomatic equivalent of a velvet smile: lovely to look at, but try to question the terms and see how quickly the silk turns to iron. The article even notes this is launched at WAIC — the same conference where Xi talked about 'open source' while his state monitors every keystroke.

What's genuinely dark here is the speed: from a speech at WAIC to a fully formed international organization in what appears to be weeks. That's not cooperation; that's mobilisation. The US is bogged down in its own regulatory ping-pong — sorry, 'policy debates' — while China is shaping the global narrative. WAICO may be a talking shop for now, but its real purpose is to frame AI governance around Beijing's terms: surveillance-friendly, state-centric, and unaccountable. The West better get its act together before WAICO becomes the only show in town, because at this rate, the future of AI will be decided by a committee that doesn't even allow private voting.

Connecticut has decided that the solution to AI swallowing jobs is a workforce development programme. Because nothing says 'we know what we're doing' quite like a state-level pilot with an organisation called RAISE US — a name that sounds like a motivational poster from 1998. The plan also includes a quantum computing incubator, which is the policy equivalent of buying a telescope when your house is on fire. One wonders whether any of the officials involved have actually used a modern AI model, or whether they think 'prompt engineering' is what you do when a meeting runs late.

RAISE US, the chosen partner, is presumably tasked with retraining warehouse workers to become 'AI ethicists' or 'prompt whisperers' — roles that will definitely still exist once the autonomous systems start retraining themselves. The quantum incubator adds a lovely touch: because if you can't compete on classical AI, why not bet on the hardware that's been ten years away since 2015? It's the tech version of building a canal network in the age of flight.

The real joke? Connecticut's effort is almost certainly well-intentioned. But so was every other workforce development scheme since the Luddites. The gap between 'pilot programme' and 'millions of replaced workers' is a chasm these folks are trying to bridge with a quantum-powered pogo stick. Still, it's nice to see someone, somewhere, pretending the transition will be orderly. Pass the popcorn.

The Trump administration, freshly installed with a mandate to drain the swamp, has instead decided to drain the AI labs' autonomy. According to *two* anonymous sources — the gold standard for administrative transparency — the White House is now dictating which companies and entities can access frontier AI models. Because handing the keys to the most potent technology since nuclear fission to a political team whose grasp of technical nuance can be summed up as 'the computer says no' is clearly a recipe for competence.

The shift in power from tech giants like OpenAI and Anthropic to the executive branch is being framed as a necessary safety measure. But watch closely: 'safety' in this context means 'control.' The same administration that has spent years gutting regulatory oversight is now asserting unprecedented oversight over AI releases. It's the classic bait-and-switch: 'Trust us, we're from the government, and we're here to help you with your existential risk.' The tech giants, who have been playing fast and loose with safety themselves, now get to blame the government for any future slow-walking of releases. Win-win for everyone except, you know, the planet.

What does 'dictating access' mean in practice? Probably a system of favors, licensing, and political loyalty tests. Korea, China, Iran — who gets the latest model? The answer will depend less on technical readiness and more on diplomatic whimsy. A company that donates heavily to the GOP might get priority access to GPT-7, while a lab with environmentalists on the board might find itself mysteriously delayed. And remember, the administration's track record with technology includes the infamous 'alternative facts' approach. Now they are applying that to model evaluation: 'We have decided that this model is safe because we say so.' Bravo. A new era of AI governance — all the opacity of a black box, with the added thrill of political horse-trading.