Anthropic to meet with White House over AI tool suspension

Executives from the artificial intelligence company Anthropic are scheduled to meet with senior officials from the White House and the US Department of Commerce on Monday in Washington D.C. The discussions aim to address fresh national security concerns surrounding Anthropic's recently released AI tools, Fable 5 and Mythos 5.

The US government had previously restricted access to these tools, specifically prohibiting foreign nationals from using them. This action followed the government becoming aware of a potential "jailbreak" vulnerability, which could allow the AI to perform unintended functions. Fable 5 is a public version with enhanced safeguards, while Mythos 5 has different controls and is restricted to select organizations. Anthropic acknowledged last week that releasing Fable 5 "comes with risks" and stated its capabilities exceed any previously released model.

Anthropic has indicated it has only received "verbal evidence" of the alleged jailbreak vulnerability. The company's CEO, Dario Amodei, is expected to attend the meeting with Commerce Secretary Howard Lutnick. A White House spokesperson declined to comment, and representatives for Commerce and Anthropic did not respond to requests for comment.

This meeting is the latest development in a series of tensions between Anthropic and the US government regarding AI usage. Earlier this year, Anthropic sued the US Department of Defense over disputes concerning the application of its AI models. However, relations appeared to improve following a "productive" meeting with senior White House officials a few weeks prior. The upcoming meeting is expected to involve more detailed documentation of the alleged security issue, though it remains uncertain whether Anthropic will regain permission to make Fable 5 and Mythos 5 publicly accessible afterward.