Anthropic Engages Trump Officials Over Fable 5 AI Export Ban
AI company Anthropic has met with Trump administration officials to discuss a federal export ban imposed on its latest artificial intelligence model, Fable 5. The ban, stemming from perceived security vulnerabilities, led to Anthropic withdrawing the model from non-U.S. public access. Anthropic officials presented their cybersecurity safeguards during the discussions, arguing that any breach was minor. This incident marks the first time the White House has compelled an AI company to remove a model from public availability.
Anthropic staff recently met with officials from the Trump administration following a federally imposed export ban on its artificial intelligence model, Fable 5. The ban, enacted due to potential security vulnerabilities, required Anthropic to prevent non-U.S. users from accessing the model, effectively forcing its removal from the market.
Anthropic has contended that any detected breach was minor and not a "jailbreak." The company sent senior leaders with expertise in research and safeguards to Washington D.C. for in-person meetings on Monday, which followed multiple hours-long calls over the preceding weekend.
Key participants in the discussions included Anthropic co-founder Tom Brown, head of public policy Sarah Heck, Logan Graham from the Frontier Red Team, head of safeguards Dave Orr, and lead security researcher Nicholas Carlini. Representing the administration were Commerce Secretary Howard Lutnick, National Cyber Director Sean Cairncross, and Chris Fall, who heads Commerce's Center for AI Standards and Innovation. Treasury Secretary Scott Bessent, previously involved, was less active in the weekend and Monday meetings.
During the meetings, Anthropic provided a presentation to administration officials detailing its cybersecurity safeguards, aiming to resolve the restrictions. A resolution is not expected within a few days, though officials left open the possibility for a quicker outcome, noting it depends on Anthropic.
This export restriction represents a significant escalation in the administration's efforts to regulate the rapidly developing AI industry. It is also the first instance where the White House has mandated a company remove an AI model from public access. Experts have cautioned that such export controls could potentially hinder U.S. leadership in AI innovation and competition.
Anthropic has defended its safeguards, stating that the identified vulnerabilities appear relatively simple and that other publicly available models can also discover them without requiring a bypass.
According to Business Insider, citing POLITICO, this development unfolds as the U.S. government implements unprecedented AI security measures.
