Loading summary
A
Today, though, we're going to start talking about the Fable 5 ban. So what's going on? The US government issued an export Control Directive citing national security concerns, suspending all access to Anthropic's Fable 5 and Mythos 5 for any foreign national that includes Anthropic's own foreign national employees, forcing Anthropic to abruptly disable both models worldwide for all customers. All other Anthropic models are currently unaffective, but Anthropic received the directive at 5:21pm Eastern previously, and the government's concern was stated to be a jailbreaking of Fable 5. So let's talk about Anthropic first. They said the demonstrated technique for jailbreaking found a small number of previously known minor vulnerabilities. Other publicly available models are able to discover them as well without requiring a bypass. And they named OpenAI's GPT 5.5. They also said, we disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people. It would essentially halt all new model deployment for all frontier model providers. Now, on the other hand, the administration's case, and this was data David Sacks said, quote, fable is Mythos with guardrails, but if those guardrails fail, then you've exposed Mythos and its advanced cyber capabilities to people who shouldn't have them. The admin asked Dario to fix the jailbreak or de deploy the model. Dario refused. Anthropic prioritized the continued offering of the consumer model over safety. And the ball is an anthropics court now acknowledging up front that we don't have a full story from either side and we're still trying to figure out what's going on. I wanted to start Sam, in particular with you because I know you've dealt with crypto and seen administrative actions previously. How are you taking this whole thing apart? What do you think here?
B
Sure. Great question, Austin, Ram and Chris, thanks for having me on. So this is an interesting we're in an administration that has been in many ways a huge friend to the digital asset sector and to the AI industry. Looking and the president has made clear he wants to remove red tape and make America the the capital for crypto, for AI, for tech in the world and and to create the kind of regulatory environment that will attract entrepreneurs to build here and disseminate their products. Right. And we are in for AI. We're in a race against China in many respects. And in the blockchain industry, there are many projects that integrate AI with crypto, integrate AI with DeFi, and regardless have some of the basic tech infrastructure that could be regulated by export control. So how do I take it? On the one hand, I think because this administration has been so friendly to the tech industry and you have the statement from David Sacks about why you can imagine that just taking a step back here, we know that Mythos is extremely powerful and was in many ways marketed as something so too hot to touch. Right. So dangerous Fable is supposed to have safeguards. And you could imagine, according to the reports, and I don't know if these are true, but what I've seen in the press is there's reporting that Amazon's CEO told the administration there was a jailbreak, in other words, a way around the safeguards. And 24 or 48 hours later you've got this 5:21pm Friday night directive not to export it. So on the one hand, you can understand as a human who may be concerned about national security, why the administration would react this way. On the other hand, as somebody who has lived through what I'll call the Howie or the Gensler wars watching, not necessarily this administration, but government officials take ambiguous regulations or ambiguous statutory authority and then use it to crush developers, innovators and businesses. It is concerning when the government has the power to effectively, effectively ban the dissemination of software, of ideas. There is well established case law that source code software, including encryption software, is literally speech, is literally protected speech under the First Amendment. Here we have not some of the things that are concerning about this, and that's not to say the government doesn't have a valid basis, but there's, there's, there's a question of what's the substantive basis and then there's a question of the process. And here we don't have, as the public, we don't have much process. The letter that was reportedly sent by the Commerce Department to Anthropic has not been publicly released. We have a description of it from Anthropic, but we don't know what it says. We don't know what the, what the concerns were, what the national security security concerns reportedly or allegedly were. We don't know what the statutory or regulatory authority for it was. And this could have, if this power is abused, it could have sweeping implications for the AI industry, but also for the crypto industry.
C
Right.
B
I mean, just taking a step back, what happened here is the regulation invoke the export control regulates the export of what? Here we're talking about APIs. Okay. Does that mean if I have a US server with an API and someone in China accesses it, that that's an export. If so, does that mean the export power can be used against any US based server, API or technology? And you know, there are exemptions or exceptions for export regulations for things that are published, which has often been why permissionless defi protocols, you don't see them getting regulated through export directives. That said, what if I have an API? What if I run a front end or, or a defi protocol that relies on an underlying AI model like Fable or something else?
A
I'll say for my part, having lived through the operation choke point side of the Biden administration as well, some of the arguments that I'm seeing here, I worry even if the people currently in charge believe them are forming a playbook to misuse them in the future. Right. Because now we live in a world where you can point at any frontier development that might have cyber security or financial stability implications, say the words national security and then start imposing things like export controls. To me the question is, where is the limiting principle? Like, one of the beautiful things about crypto is you meet an incredible amount of interesting people working in the space. And one of the ones I've met is Gene Hoffman. And if anybody remembers the original encryption wars, Gene was at PGP at the time that was happening. And was the guy carrying around a physical book of printed code, right. To get around export controls because the First Amendment allowed him to do that. And I think it is a fascinating thing to see all of these principles that we've been fighting about now for 20, 30 years coming back front and center with this discussion, because what I will kind of raise here, and Chris, you've like lived through some of the national security side of things. Those concerns are not invalid. There are genuinely valid national security concerns if people could use these tools to damage the United States. But how do you contextualize those? Like, where would you put a dividing line? Because you're also somebody who's operating.
C
National security will always trump responsible innovation, no matter what. And if you look right now, a lot of the vectors of attack you're seeing in Congress right now on the Clarity act, you know, maybe banks are so happy about interest anymore and maybe there's a different way to kill this thing by going after national security. Right? So that is always the trump card, no pun intended. As you look at this right now, I was at this event this morning and I spoke and afterwards Hillary Clinton came on and the one thing that jumped out at me, and look, she was Very articulate about her perspective around Iran. And maybe we'll talk about that later. But at the end, the moderator said, hey, what keeps you awake at night? And she said, you know what really keeps me awake at night? AI. And so if you think it's bad right now with the government intervening in anthropic, and I think we could have a long debate about whether this was politically motivated or whether it was not. And some people say, oh, they just hate Dario, so they're going after him. But I think we're just getting warmed up and I think AI has a pretty big problem right now that crypto got in front of. Frankly, you know, maybe our PR isn't the best, but we were, we had a very good game in D.C. but I think both these technologies have a PR problem. And one thing, as we're going into elections, midterms, you know, I saw a lot of Democrats get on stage this morning and talk, and they emphasize moderation, moderation, moderation. But then Hillary came up and said, AI keeps me awake at night. So I think we're getting warmed up. I love the fact that you mentioned the encryption wars because we continue to bump up against this every time there's a new technology. And over time we figure it out. The problem is, what do we do? You have the Chinese that are coming up with very sophisticated models, very cheap models. The more and more we get involved in and put our hand government and over regulate it, a couple things got to happen. Number one, consolidation, right? We see it over and over again. Regulation leads to consolidation. It boxes out entrepreneurs because it's very expensive. And you listen to the all in guys, they're like, hey, that's the playbook here. Get big enough to create that regulatory moat. And that's been the game, that's been the game for a very long time. And so that's the risk that we face. I don't think you can put this toothpaste back in the tube. And then it brings you back to this whole idea of decentralized AI. And we've been in this space, you know, for a very long time, very, very hard, unproven. And there are parts of the stack that you can decentralize. But generally you always have somebody centralized, no matter what, like whether it's that front end or something else, they're going to be subject to regulation. So maybe the future is like the AI mullet, where you have the decentralized AI in the back end. I don't know. But I think this is going to get much more Complex. I think it's only going to get worse because if the Trump administration is already taking this much of an active viewpoint, then I think the Democrats get it. Last thing is that we're in this period right now of max concern because these models are moving faster than our security. Eventually I think we're going to reach the stasis where the models actually fix the security. And once your vulnerabilities are fixed, maybe you get some very sophisticated ones that pop up in the future. But we should end up with a much stronger environment from a security perspective once we've had the time to implement these models. So hopefully that plays out in the future.
D
Yeah. Nodding my head at the comments. I mean, overall the limiting principle is the US Constitution, specifically the First Amendment. To Sam's point, this is an overreach of the federal government. It's disappointing. I know. Growing up I read Ray Bradbury's Fahrenheit 451 which was a lesson on the concerns around burning books. We read about George Orwell, 1984. There's Kerr Vonnegut's short story Harrison Bergeron. It's so surprising to see this come from this administration. And in the future we're not going to have our kids read books. They're going to have access to intelligence on demand like utility. It doesn't make sense for the role of government to restrict that. There's no due process. National security has been used and abused to contravene on natural rights and it's being abused here. I agree. Also, AI is becoming a political issue. I had a chat this Sunday as my Verizon fios went down. I call into get it repaired and I asked the service operator, oh like where are you based? Just making some chit chat while the IP is updating. She said, I'm in Virginia, Virginia Beach. I said that sounds lovely. She said, no it's not. They're building a data center here. And I said, oh well, it's just a one time construction issue. She said no, no, no, no. It's already up and running. It's worrying, it's making sound. My utility bill is going up, it's having environmental damage. Her words. So this will be a continued issue. You know, AI is important for unlocking the potential of humanity and you need political leaders at this time to step up and communicate that story. What's more likely to happen though is that AI will become a punching bag and used to create the next enemy or scapegoat. So yeah, I agree, you'll see more backlash And I think Anthropic is betting on a change in the midterms. They have a bunch of ex Biden staffers is my understanding too. That dynamic is in the background.
A
How much of this though? Right, because like OpenAI clearly caught a stray in Anthropic's commentary on this and yet nothing has been done with them. We haven't seen anything coming out about Grok or any of these sorts of things. How much of this is related to what Sam brought up earlier, which was Anthropic's own marketing of its own models? Because if you run around saying my model is an existential threat to cybersecurity, whose fault is it if people take you at face value?
D
They have the best model. They have the best model on the block. But I think, yeah, Chris is right. It's not the proper role of government here to set cybersecurity policy. It's company IP property and private enterprises very good at identifying issues and risks of their business and innovating new solutions to address it. Like Y2K should have been a bigger issue than this. You didn't have national security telling private companies to go address this Y2K bug. Private companies identified the issue and they invested in technology and it became a non issue. And that should happen here.
C
It's very.
B
Oh, sorry Sam.
C
No, you go saying go.
B
I was going to say that. I think the problem here, right, Is if in fact this type of. If you really are concerned about national security, you can't stop Chinese models, decentralized models, or anyone who was able to crib the mythos or fable tech in the period it was out deploy it. Right? So this isn't really, if you're really truly trying to solve for national security, this is not an effective solution. But you are disabling the market from getting up to speed on it, to fix it. And if we believe in the free market, then we should let the free market toy around with this.
C
What makes me crazy, people get freaked out about tech. They're naturally scared of it. And the immediate response is let's go after the tech, let's regulate the tech, let's attack the tech, right? But the point is, is that it's not the tech that does anything. It's the users, it's the people, it's the institutions and it's the governments that use it. Like maybe we should get rid of cell phones because you can do a lot of bad things on a cell phone. And I thought about this all the time when I was in Iraq, I used to get. We would get blown up by garage door openers. They trigger IDs on us. So, like, garage door openers, let's get rid of them. They're dangerous technology. Right? So, like, I wish we would focus more on the behaviors and hold people, the front ends, accountable rather than just go after. Like, if you use Mythos to hack somebody, well, let's go after those people and those institutions and hold them accountable like that. I feel like we always default to the tech, and that's the wrong place to default.
A
I feel like it's also the lazy path in some ways, because people, and by the way, back to Sam's earlier parallel with crypto, people have this problem with the Internet now where it's global. And so you naturally want to expand your reach to regulate things, but some of those people might be outside of your specific national borders. I don't think anybody's really come up with a coherent, call it theory of regulation in the Internet era where we could effectively address those things without also stepping all over extraterritoriality. And I feel like, as a result, they reach for the lazy solution. Like, Sam, you do a lot of legal work in this space. I guess I would ask you, if you were advising the US Policymakers thinking about this right now, of how to balance these concerns, like, and I'm going to put you in the awful spot of thinking about having to make this decision yourself, how would you be approaching this right now as the US Government?
B
So I think the first thing is they have to communicate their reasoning. If it is, in fact a valid national security concern. It is very important. There are mechanisms to regulate through notice and comment rulemaking, which is the bare bones of democracy. Right? The bare bones of due process. This goes around that, that's extreme, fine. But lay out what is the concern? What is it? What is your statutory authority to do it? That, at a bare minimum, is very important so that all of us can understand. What is the framework here? When President Trump came in, he rescinded, there had been an AI diffusion framework set up by the Biden administration for how the export regulations would apply to AI models. And he scrapped it. Okay, but we still don't have a replacement. Right. So what is the framework here? What is the precedent being set here? If you want the AI application makers like Anthropic OpenAI to comply, to take safety seriously, to have those guardrails, you need to set the rules of the road. And that's not my own message. That's the Trump administration's message. Right. The complaint against Biden and Gensler was regulation by enforcement, not guidance. Okay, well that approach approach needs to be taken here.
A
I think that's actually quite fair in terms of say what you mean and write it down. In many ways it echoes the complaints about like secret proceedings at visa courts. Right. And privacy violations. There is you have all these undisclosed things that we never know about. You took a bunch of actions and we have no way to know if they were completely valid and totally reasonable or complete nonsense and violating the rights of Americans. So I guess step one is transparency and some later going to be the ultimate disinfectant here for sure.
B
It's hard to go beyond that because we don't have the facts. We have Anthropic's version. I'm not saying they're wrong, I just don't know. You've got the SAC statement which gives you a window, but that's not sufficient data for me to make an assessment of. What exactly is the rule? How does this rule apply to the AI industry, to the DEFI sector, to blockchain? All of these are unanswered questions. With something this dramatic, we shouldn't have those questions.
A
If you like this segment, please like subscribe and tune in every Monday at 4:30pm Eastern Time. I'm Austin Campbell, the host of Bips and Bips or, along with my friends Rahm Alawalia and Chris Perkins and our slate of exceptional guests. Every week we're going to discuss macro, crypto and the collision of worlds, covering topics that move markets and shape the financial landscape.
E
Fidelity has been researching and investing in blockchain since 2014, long before it was a headline, and they're hiring crypto and DEFI professionals to join their team and discover what's next in finance. Fidelity is looking for people with fresh perspectives from different backgrounds, whether it's tech, UX or product design, Whether you're crypto savvy or crypto curious, as long as you have the passion to make a real impact at a company striving to make finance accessible to all. Explore crypto careers at Fidelity today and make the decision that could change your future for the better. Visit crypto.fidelitycareers.com to learn more. That's crypto.fidelitycareers. com Fidelity is an equal opportunity employer.
Host: Laura Shin
Air Date: June 18, 2026
Guests: Sam Enzer, Chris Perkins, Austin Campbell, Rahm Alawalia (rotating voices, see below)
Topic: The US government's sudden export ban of Anthropic’s Fable 5 AI model, its broader implications for AI, crypto, and free speech
This episode explores the recent US government ban on Anthropic's Fable 5 and Mythos 5 AI models, citing national security concerns. Host Laura Shin leads a roundtable featuring legal and crypto experts (notably Sam Enzer and Chris Perkins) to dissect why this administrative move sets a potentially troubling precedent—not just for AI, but also for the evolving regulatory landscape around crypto, DeFi, and open-source technology.
The panel discusses the practical, legal, and philosophical impacts of treating AI models as "exportable" commodities and the chilling effect of ambiguous government intervention—including free speech, due process, and the dangers of regulation by fiat. Parallels to the 1990s encryption wars and previous government crackdowns on crypto innovation abound, along with signals about what may come next for decentralized tech.
[00:00–02:00]
[02:09–05:38] - Sam Enzer’s Legal Perspective
[05:39–08:15]
[08:15–11:55] - Chris Perkins’ Perspective
[11:55–13:55] - Rahm Alawalia’s Perspective
[13:55–15:08]
[15:08–16:51]
[16:51–17:49]
[17:49–19:48] - Sam Enzer’s Prescriptions
[19:17–19:48]
This episode of Unchained delivers a rich, urgent discussion of how the US’s ban on Anthropic’s frontier AI models signals a new chapter in the ongoing tension between tech innovation and regulatory intervention. Echoing the encryption and crypto “wars” of past decades, the guests argue that ambiguous, unaccountable, and sweeping government actions threaten not just AI but the foundational freedoms that undergird technology, industry, and expression. The panel calls for transparency, process, and clear legal frameworks to avoid stifling technological progress under ill-defined claims of national security.