White House Tightens Export Controls on Anthropic Over AI Safety Fears

The pressure mounted quickly after Amazon CEO Andy Jassy reached out to the White House on Thursday, just 48 hours after Fable’s release. Jassy, whose company maintains a significant investment in Anthropic, flagged risks that the model’s safety guardrails could be bypassed. Following this, an array of senior officials—including Cyber Director Sean Cairncross and Deputy Chief of Staff Richard Walters—engaged in a high-stakes negotiation to force a voluntary removal of the software.

Anthropic CEO Dario Amodei resisted, contending that the identified security gaps were specific rather than a total system compromise. Despite these assurances, the administration issued a strict 90-minute deadline to pull the model. While Anthropic later defended Fable in a blog post, claiming that no “universal jailbreak” exists and that their safeguards are sufficiently robust, the government viewed the intervention as a last resort. An anonymous official characterized the move as a forced hand, noting that earlier efforts to resolve the security dispute through cooperation had failed.