It gets weirder. The Claude ban just got a part two.

First — the tweet that started it all.

If you missed part one, here's the post that lit the fuse. Anthropic announced the ban on X themselves. Read it, then I'll show you everything that's happened since.

That's the official line. Models suspended for every foreign national, worldwide, including Anthropic's own staff. Now here's what nobody saw coming.

A rival lab turned Anthropic in.New

This is the biggest new piece. The "jailbreak" that triggered the ban didn't come from a random hacker. It came from researchers at Amazon — and they reported it straight to the US government.

Let that sink in. Commerce Secretary Howard Lutnick sent the letter directly to Anthropic's CEO, Dario Amodei. The trigger was another company demonstrating the jailbreak — and the wild part is, the bar was incredibly low.

So one company found a soft spot, told the government, and the most powerful AI on the planet went dark for everyone. That's the part the headlines buried.

The famous jailbreaker went public.New

Pliny the Liberator — one of the most well-known names in AI jailbreaking — posted his own break of Fable 5. But the part worth hearing is his verdict on the whole release.

His point matters. The safety lockdown didn't just stop bad actors. It blocked the legitimate security researchers who could have made the model safer. That's the irony running through this whole story — the harder Anthropic locked it down, the more it frustrated the exact people it needed.

Even Andrej Karpathy just got locked out.New

Here's where the ban stops being abstract. "Any foreign national" includes some of the most important AI researchers alive. Andrew Curran pointed out a perfect example.

One of the people who helped build modern AI — barred from the model by the clock, at 5:21pm, because of his visa status. Multiply that across thousands of brilliant engineers in US labs who happen not to hold a passport. That's the real cost of a rule written this broadly.

A black market just lit up.New

Ban something powerful and you don't make it disappear. You make it valuable. Zhu Liang put a number on it.

There's already a whole industry reselling Claude access into markets that can't get it directly. The ban doesn't close that door — it pours rocket fuel on it. Restrict the official supply, and the unofficial supply explodes.

The contrarian take — this is a marketing win.New

While everyone panicked, a few people pointed out the obvious. What's the ultimate flex for an AI company? Building a model so capable the government had to switch it off.

And there's a sharper version of this take — that Anthropic, the company most vocal about AI safety regulation, basically got the outcome it has been campaigning for.

Whether you buy that or not, the branding is undeniable. "The model so smart it got banned" is a headline money can't buy.

The backlash got loud.New

Not everyone is impressed. The reaction split into three camps, and all three are worth hearing.

Camp one — "this is government panic"

Camp two — "this is gatekeeping intelligence"

Camp three — "we just lost something special"

That last one is the quiet fear under all the noise. People got a taste of the most capable model ever built — and it was taken away in a single evening. The meme version of that feeling spread fast too.

Where this goes next — ID to use AI.New

Here's the development that should actually shape how you plan. Once a model can be restricted by nationality, the obvious next step is proving who you are. Peter Yang called it.

And there's a deeper twist the reporters caught. Anthropic spent two years warning everyone how dangerous powerful AI could be. This week, those exact warnings were used as the reason to switch its own model off. The safety pitch became the kill switch.

So this is the direction of travel. Access to the best models is going to get more gated, more political, and more tied to who you are and where you live. Which leads to the only question that matters for you.

The ban itself, in four lines.

That's part one. If you want the full breakdown — the timeline, Anthropic's full statement, and every source — watch the first video. This guide is about everything that's happened since.

The model is not the moat. The system is.

Everything in this story points to one truth. The best models are going to come and go faster than ever — banned, gated, ID-locked, rate-limited. The people who win aren't the ones chasing each new model. They're the ones who own the system around it.

Every source, first-hand.

Everything here came from the people who broke it on X and the outlets that confirmed it. Go check it yourself.

Everything new — at a glance.