The U.S. government has recently lifted export controls on Anthropic’s advanced AI models amidst ongoing discussions about regulatory practices in the artificial intelligence sector. In a statement issued by Anthropic on Tuesday evening, the company announced that the Department of Commerce has removed export restrictions on its models, Claude Fable 5 and Mythos 5. The company plans to restore access to these models starting immediately.
This announcement followed a post by Commerce Secretary Howard Lutnick, who expressed satisfaction over the decision. He noted, “Over the past two weeks, we have worked closely with Anthropic to analyze and approve Fable 5 to ensure alignment across the U.S. Government and strengthen America’s leadership in AI.” His statement underscores the collaborative efforts between the government and Anthropic to align on safety and security protocols.
Claude Fable 5, part of Anthropic’s advanced AI suite, is essentially a version of their Mythos model, enhanced with additional safeguards to ensure its suitability for public and commercial use. The decision to impose an export ban earlier this month was a response to concerns raised after a trusted partner, identified as Amazon, discovered vulnerabilities that could potentially allow users to bypass Fable’s regulatory guardrails. At that time, Anthropic acknowledged that the jailbreak methods were straightforward and admitted that similar vulnerabilities existed in other publicly available models.
As a direct consequence of these vulnerabilities, the Commerce Department mandated that Anthropic suspend all access to Mythos and Fable for foreign nationals, even affecting Anthropic employees. This led the company to temporarily disable access to both AI models.
Following further negotiations and assessments of the cybersecurity risks, the Commerce Department had granted Anthropic permission to release the Mythos model to selected, government-approved entities. However, neither Anthropic nor the Commerce Department disclosed the specific modifications made to Fable in order to have the export ban lifted.
Experts have raised concerns regarding the capabilities of Mythos, citing its potential to exploit cybersecurity vulnerabilities at a rapid pace. Due to these potential risks, Anthropic has limited the model’s release to a select group of key partners deemed essential for securing critical software infrastructures.
Additionally, the White House has requested OpenAI to restrict the release of its upcoming GPT 5.6 model to a small number of government-approved partners, citing similar concerns about advanced AI capabilities. OpenAI has voiced its apprehensions regarding the establishment of such government access processes as a long-term standard within the industry.
As the dialogue around AI regulation continues to evolve, the recent developments between Anthropic and the U.S. government highlight the delicate balance between fostering innovation and ensuring safety in the rapidly advancing field of artificial intelligence.
