US Orders Anthropic to Disable Foreign Access to Mythos artwork

US Orders Anthropic to Disable Foreign Access to Mythos

Bloomberg Businessweek

June 15, 2026

The people, companies and trends shaping the global economy. Watch Carol and Tim LIVE every day on YouTube: http://bit.ly/3vTiACF.
Speakers: Tim Stenovec, Maggie Eastland, Carol Massar, Bailey Lipschultz, David Bauer, Geetha Ranganathan, Carmen Reinicke
**SPEAKER_1** (0:02)
Bloomberg Audio Studios, Podcasts, Radio, News.
This is Bloomberg Businessweek Daily, reporting from the magazine that helps global leaders stay ahead, with insight on the people, companies, and trends shaping today's complex economy, plus global business, finance, and tech news as it happens. The Bloomberg Businessweek Daily podcast with Carol Massar and Tim Stenovec on Bloomberg Radio.

**Tim Stenovec** (0:32)
Just a warning to everybody, we're going to get a little technical, but Maggie's going to help us understand what jailbreaking is and what's at stake when it comes to anthropic, fable 5, and mythos. Anthropic executives planning to meet with Trump administration officials on Monday to discuss an unprecedented US government directive barring foreign access to the company's most advanced AI models. That's according to a person familiar with the matter. For more, we're joined by Maggie Eastland, Bloomberg News tech reporter. She's based in Washington, DC. It's where she joins us from right now. So Maggie, I said this would get a little technical because I think we should just start off by explaining the Fable 5's relationship to mythos and what has the government right now so concerned and why Anthropic is pushing back on this.

**Maggie Eastland** (1:15)
Yeah, absolutely. It definitely helps to sort of get some definitions down to start. So mythos, that was released in April by Anthropic. It's their latest model, but it was not released to the public because the company was too worried that it had a lot of cyber capabilities and it could be used for cyber attacks. So it was only released to a limited group of government and business partners to sort of help those partners shore up their own cybersecurity efforts. What happened last week was Anthropic actually released a version of that model with some added safeguards to the public. So that's what we call Fable 5 and that's sort of what started this whole situation between the government and Anthropic. A few days after that public release, the government sent Anthropic a letter that was an export control directive that essentially blocks anyone who was a foreign national from even accessing that Fable 5 or that Mythos 5 model. So in effect, Anthropic was forced to immediately disable those models, especially since they have foreign nationals who are employees. So this affects their employees, it affects their customers, and all of a sudden, they're only able to provide these models to US citizens. And they don't even have a way currently of sort of a know your customer standard in terms of where those people reside who are using their models.

**Tim Stenovec** (2:37)
I'm glad you said that because I'm opening, Claude, from Anthropic right now on my phone. I don't pay for it, but I have it, I can use it.
I had the ability to use Fable 5 if I paid for it, but now it says currently unavailable. It also advertises that it's for your toughest challenges. But Claude has no idea where I live in terms of what my nationality is. So they just have to disable it because they don't have those Know Your Customer guardrails in place.

**Maggie Eastland** (3:06)
Exactly.
It would definitely take some time to develop those. There could be some privacy concerns from folks in Washington. It certainly wasn't something that they could do on very short notice.
And they were sent this letter at a pretty quick turnaround. So, you know, as we've reported, Anthropic executives are now in Washington, DC. They're meeting with the Commerce Department today. We'll see what comes of that. They're trying to work through this. The concern here, I should point out, and Anthropic has said this in their statement, they believe that the government is concerned about a jailbreak. Now, there's not necessarily one definition for what that is, but typically it means that users who are using Fable might be able to prompt it in a certain way so that they're able to access those cyber advanced capabilities that they're not supposed to access. So Microsoft was involved in surfacing this vulnerability, and that's certainly going to be dominating the conversation in terms of just trying to figure out if this is something Anthropic could fix, if they could potentially fix it with more of these know-your-customer standards. So that's what's going on right now, and we'll have to see if they're able to work this out and somehow remove that export control.

**Carol Massar** (4:22)
So on one hand, Maggie, it seems like the White House, the Trump administration is actually very keen to what could possibly happen and is staying ahead of the game, it almost sounds like. Who is leading the strategy and the process when it comes to the White House? And it sounds like the expertise is there. Is that correct? Because I feel like with social media, we felt like maybe everybody didn't quite understand everything that was going on.

33 more minutes of transcript below

Feed this to your agent

Try it now — copy, paste, done:

curl -H "x-api-key: pt_demo" \
  https://spoken.md/transcripts/1000651996090

Works with Claude, ChatGPT, Cursor, and any agent that makes HTTP calls.

From $0.10 per transcript. No subscription. Credits never expire.

Using your own key:

curl -H "x-api-key: YOUR_KEY" \
  https://spoken.md/transcripts/1000772867501