Anthropic staff were set to meet senior White House officials Monday after the US government restricted access to the firm’s most advanced AI models.
The Trump administration blocked foreign nationals’ use of Fable and Mythos, citing security concerns: The models are highly cyber-capable, and Anthropic only released Mythos to select organizations to plug security vulnerabilities.
But the move comes against a backdrop of enmity, after the Pentagon branded Anthropic a “supply chain risk” over the latter’s refusal to allow its models to be used in fully autonomous weapons.
Ironically, the US government’s restrictions could help cyberattackers; open-source models are increasingly powerful, and since hackers tend to use those, limiting frontier AI access will hurt institutions’ ability to build defenses, an open-source AI platform argued.