Decoding the battle between the White House and Anthropy over the myth

Hello and welcome organizeremail to edge Subscribers about technology, politics, and what happens when science and self-interest collide. Not a subscriber? Register here today! Got the scoop on a trivial dispute that will somehow reshape the entire field of frontier AI development? Send them to tina.nguyen+tips@theverge.com.

Back when I was covering Donald TrumpDuring the first US presidency, it was incredibly common to read three different versions of the same story. His administration has been divided into several factions, each with different interests, and all… Hated each other. There was Reince Priebus The traditional Republican Party faction, based in the Manhattan community Jared Kushner Primitive populist faction Steve Bannon Faction, deep state John Kelly Conspiracy faction-MAGA Mike Lindell faction, and so on. Over time, you could identify the camp that was leaking narratives to the media, either to undermine its rivals or to save its own reputation. In fact, for decades, media manipulation has been a popular survival tactic in Trump’s world, which often relied on factions and fierce competition for Trump’s approval. like Associated Press I mentioned On November 26, 2016 In an article about his pre-presidential business style, “[a]ides often pitch him suggestions via the media, knowing that Trump is a voracious watcher of cable TV and may be convinced by what he sees and hears.”

This penchant for backbiting and cross-purposes once again gives us a little glimpse into the White House, this time explaining some of the chaos surrounding the Trump administration’s decision Friday night to impose licensing restrictions on Anthropic’s advanced Fable model. Even after several days of statements, reports, and negotiations, it remains unclear what actually happened behind the scenes, and even less clear who was responsible for what could have been a massive cybersecurity disaster.

To recap for the uninitiated: On Friday evening, the White House imposed export restrictions on Anthropic’s Mythos 5 and Fable 5, their most recently released models, preventing foreign governments and citizens from using those products. It essentially forced Anthropic to shut down access to both models entirely, throwing its user base into disarray, putting the company’s future in doubt, and the overall prospects for frontier AI in a gray area. Can the government simply tell AI companies to go out of business? Over the weekend, several conflicting accounts emerged, though some points remained consistent: White House allies claim that within days of the launch, several tech executives were gutted, including Amazon’s president and CEO. Andy Jassyhave been contacted with concerns that Fable and Mythos could be jailbroken, posing an imminent threat to national cybersecurity. The two parties spoke on Friday, but the nature and duration of their contacts varied according to the account: The Washington PostAn account has been claimed Anthropic was given 90 minutes to remove their models, while… A White House official told Politico They pleaded with Anthropy “for hours.”

From here, the details become murkier, at least from the White House’s perspective. says one party the New York Times Amazon has found a way to “jailbreak” the security barriers that prevent users from using Fable for cyberattacks. A second party responded times One can achieve the same results using ChatGPT 5.5 from OpenAI. Semaphore reported that it maybe This was due to a China-linked group gaining access to Mythos, though the actual jailbreak has not been confirmed. and Axios sources indicated The administrator simply did not like the woke atmosphere at Anthropic. “Anthropic has not done a great job of trying to talk to the administration and appreciate ideological differences,” a source familiar with the administration’s thinking told Axios. “It’s like they’re speaking different languages.”

EdgeChief Artificial Intelligence Correspondent Haydenfield He has A more comprehensive report here Based on the view of Anthropic and the AI ​​safety community, which says concerns about jailbreaks are overblown. But inside Washington, the overwhelming consensus I hear is that even if they are ultimately right, they are humane and… Dario Amodei They have created their own political nightmare by opposing the Trump administration — and not in the morally superior way of resistance. “They don’t bend and Dario is stubborn and says what he thinks even when it’s stupid and they have (whether justified or not) a holier-than-thou feeling,” one AI policy advocate pointed out to me.

One of the most interesting topics that came up in organizerOur trajectory is how often the hard logic and irrefutable facts rooted in science and technology collide with the emotion-based corruption of politics. In the absence of any federal regulatory law, regulation of artificial intelligence He is Relying entirely on the reactions to whatever statement someone convinced Donald Trump to post on Truth Social, or the executive order he was persuaded to sign. Unfortunately, as a former White House AI advisor Dean Paul As he pointed out in his Substack on Tuesday, this is the reality under which frontier AI companies must operate. As he writes In an inflammatory post:

What the law says doesn’t matter. It doesn’t matter what administration officials argue in a single day. Anthropologists are a political enemy of this administration, in part because they have explicitly chosen to make themselves a political enemy. It is simply naive to believe that your company can operate in such conditions without an extreme degree of regulatory caution. In this context, Anthropic’s actions are viewed by many inside Washington as not simply unwise, but deeply hostile.

And it’s not just about human and political grudge matches with the Trump administration. Everyone at the border needs to understand that as a practical matter, you need an explicit green light from the government now.

This slogan certainly applies to all companies operating during the Trump administration, especially companies working in the field of emerging technologies. But for some reason, while their competitors adapted quickly, Anthropic’s sentiments were clearly out of sync with those in the administration — and clearly no factions in the administration were rushing to defend Anthropic. Which should be raised some kind of question.

shotTwo weeks ago, before the New York Knicks and San Antonio Spurs faced off in the NBA Finals, Sesame Street Elmo He betrays his hometown.

pursuer: The Knicks won five games, but the fanbase didn’t forget.