**Julian Goldie** (0:00)
Claude, Fable 5 and Mythos 5 have both been banned by the US government. Now, I've already covered this today. But what I want to show you today is exactly how deep the rabbit hole goes, because there's loads of interesting new stuff that's just come out about this whole situation, which makes it get weirder. So this is the original tweet, if you've not seen it already, and the announcement that the US government citing national security authorities has issued an export control directive to suspend Fable 5 and Mythos 5 That was for all foreign nationals inside or outside the US, but actually what they decided to do in the end, just pull it for everyone, right? So you can't use Claude Fable 5 at all now. However, it gets stranger than that. Now, what happened after that is that a new development came out, all right? So if you're not familiar with this, basically a new development came out, which turned out the Arrival Lab potentially has turned Anthropogin. This is what's being reported, and this is a big news piece. So the supposed jailbreak that triggered the ban didn't come from like a random person. Apparently, according to the Wall Street Journal, it came from Amazon researchers. Obviously, Amazon, they create their own AI models. They're kind of like Arrival, but they also support cloud-based models as well, like Claude, and they reported it straight to the US government, according to what's being reported here, according to the sources. So according to this, you see the International Cyber Digest, and I think they were actually referencing Wall Street Journal who have reported it. Amazon researchers went to the US government about jailbreaking Fable 5 and Mythos 5, forcing Anthropic to immediately shut down Worldwide Access Security Export Control Directive, obviously got issued by Howard Lutnick after that to enforce the action. Now, if you look at the quote here, I think this is from Wall Street Journal. We'll come on to that in a second so you can see the claim about the Amazon researchers has come from this article, which is pretty crazy. And so you might be wondering, okay, what happened? Well, basically, the jailbreak research in question was done by researchers at Amazon who use a series of prompts to get Anthropics model to provide them with information about a handful of security vulnerabilities, says, Katie Mussoris, Chief Executive with the cybersecurity firm, Anthropics shared a copy of the report with her, she said. Which is pretty wild, right? So this is all apparently come from Amazon.
So if you think about this, this letter was sent directly, rival lab, not a bad actor, evidence was verbal.
Apparently, it was a pretty mundane thing.
There was obviously no proven harm or anything like that. And then it went public as well. Now, also, another interesting thing here is that there's this, this guy or girl, who basically announced that they actually jail broke Amazon as well, right? Sorry, not Amazon, Claude, right? So you can see the details here in terms of what they did and what they announced, etc. Which is pretty wild. So I've got 2 million views, and I think that will be going viral right now with people sharing it on X, etc.
But basically, what this means is that they also announced that they supposedly liberated Fable 5, which is pretty crazy in itself, right?
So this hasn't been linked to the reason why it got shut down. The Amazon research report is linked to why, but you can see multiple different situations where the model has been jailbroken. So that's pretty crazy in itself. Now, foreign nationals. This applies to foreign nationals. Who's a foreign national? Well, apparently, this is another tweet by Andrew Curran, sorry, Andrew Curran, who announced that according to GROK, Andrea Capofe is an EB1 extraordinary ability green card recipient, not a US citizen, which means that Andrea Capofe is on the Anthropic team, got locked out this as well. I don't know if these are true. This is just speculation at this point, but this is pretty interesting. So Andrea Capofe or Capofe read a site in this as well, that is an EB1, so a Slovak Canadian AI researcher, he has citizenship in Slovakia and Canada. So that would suggest that Andrea Capofe is godfather of AI, on the Anthropic team, wouldn't be able to access this because this applies to any foreign national. How crazy is that? So even like people at the top, top, top tiers, like seeing all the crazy stuff behind the scenes with Claude, even he can't access it. So if that's true, because of his visa status. Now, you multiply that across thousands of brilliant engineers in the US labs as well, who happen not to hold a US passport. That's the real cost of this is like fantastic minds, fantastic engineers who are moving this technology forward, can't access it too because they're foreign nationals, and the ban applies to foreign nationals inside or outside the US. Also, something interesting as well, you know, Julian was talking about this.
9 more minutes of transcript below
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/1000772687642