The White House ordered Anthropic to restrict access to its AI models Fable and Mythos due to national security concerns. The company was given a 90-minute window to limit access, which it did by pulling the plug on both models. This move has raised questions about the effectiveness of export controls in containing frontier AI.
Anthropic's models were previously available only to around 150 vetted companies and government organizations. However, after granting access to a South Korean telecom through its limited partner program, U.S. officials grew concerned about potential ties to China. Amazon CEO Andy Jassy also alerted the administration after his company's researchers found a way around Fable 5's safeguards.
The Commerce Department issued an export control directive, which has sparked debate over the government's ability to use export controls to contain AI. The episode is seen as a test of whether the U.S. can use export controls to contain frontier AI like it did with encryption and spyware in the past. However, previous attempts have had mixed results, and some critics argue that export controls may not be an effective approach to containing malicious actors.
Anthropic's access to foreign markets could be affected by the outcome of this standoff, and the rulebook for other AI labs may also be shaped by how it is resolved.
