The Trump administration is preparing to restore access to Anthropic's Fable 5 model, which went dark two and a half weeks ago due to national security concerns. Insiders expect the government's restrictions could be lifted by early next week, signaling a dramatic shift in what has been a tense four-month clash between the White House and the AI company.
The potential reversal came into focus Friday when Commerce Secretary Howard Lutnick authorized limited restoration of Mythos 5, Anthropic's advanced cybersecurity model. In his letter to the company, Lutnick credited Anthropic for working constructively with federal agencies to "address risks associated with" both systems. The move marks a notable thaw in hostilities that peaked in June when Defense Secretary Pete Hegseth labeled Anthropic a "Supply-Chain Risk to National Security."
Fable 5's sudden disappearance on June 12 shocked developers and companies that had grown attached to the model in its first three days of public availability. Stripe used the system to overhaul a 50-million-line codebase in a single day, a task that would have consumed more than two months of manual engineering work. The model had been billed by Anthropic as the most capable system ever released to the general public, and one industry newsletter called it "the best coding model in the world" before it vanished.
The blackout froze automated workflows across the developer ecosystem. Companies scrambled to switch to competing alternatives, including lower-cost models from China. The outage was unprecedented for a top-tier model that was already in users' hands and actively being deployed in production environments.
Two senior administration officials have played key roles in reducing tensions, according to inside sources. Commerce Secretary Lutnick and Treasury Secretary Scott Bessent both helped broker a path toward resolution after the Hegseth dispute threatened to escalate further. Treasury and other government agencies have already signaled that Fable 5 is safe to return to public access, but the Pentagon and National Security Agency have yet to grant final approval, leaving room for unexpected complications.
The original Fable 5 rollout included a limited-time offer of free access through June 22 for subscribers to Anthropic's paid Claude plans. It remains unclear whether returning users will receive the full benefits they were promised or whether access will be subject to new restrictions, additional fees, or identity verification requirements.
The standoff between Anthropic and the Trump administration reflects broader tension over how the government should oversee the release of cutting-edge AI models. Both Anthropic and OpenAI have pressed the administration to establish a clear, transparent statutory process for reviewing new systems. The current case-by-case approach frustrates the industry, which argues that ad hoc decisions keep powerful tools away from the developers, enterprises, and cybersecurity experts who need them most.
Anthropic specifically called for "a process that is transparent, fair, clear, and grounded in technical facts" when it suspended Fable 5 and Mythos 5 access on June 12. When OpenAI received permission Friday to launch a limited preview of its GPT-5.6 model, the company echoed that complaint, warning that making government approval the default could lock the most advanced tools away from users indefinitely.
The Trump administration's June 2 executive order created a framework for voluntary government vetting of the most powerful new AI models. Neither Anthropic nor OpenAI is satisfied with the results so far. Both companies argue that clearer rules and faster decision-making are essential to keeping innovation on track while addressing legitimate security concerns.
Author James Rodriguez: "If Fable 5 returns this week without permanent restrictions, it won't erase the damage to trust, but it will confirm that backroom negotiations, not bureaucratic theatrics, are how the Trump administration actually resolves these fights."
Comments