Anthropic Employees Accuse Trump Administration of Targeting Them
TL;DR
Anthropic employees reportedly accuse the Trump administration of singling out the company after the US moved to restrict access to its new Fable 5 and Mythos 5 models. The stated trigger was national-security concern over cyber capabilities and possible jailbreaks. Anthropic appears to have disabled access broadly because filtering users by citizenship is operationally messy and risky.
Nauti's Take
When model access becomes a geopolitical loyalty test, product planning turns into foreign policy. Builders should worry about more than jailbreaks: if your roadmap sits on one frontier lab, you are building on infrastructure that can be politically switched off.
Briefingshow
The episode shows how quickly frontier models can shift from product launches to geopolitical assets. If Washington can sideline models without transparent standards, AI companies, customers, and allies face a new planning risk. The safety concern is still real: models with strong cyber capabilities need verifiable rules, not just company assurances.
Sources