-1.png?w=1024)
Washington's effort to lock down access to Anthropic's Fable 5 is meeting resistance, with the company and dozens of cybersecurity leaders questioning the government's security concerns.
After US officials on Friday imposed restrictions on foreign access to Fable 5, citing national security concerns, Anthropic shut down the model. But days later, the company met with government officials in Washington seeking to reverse or narrow the measures.
Anthropic’s effort is backed by dozens of cybersecurity experts who’ve signed a letter urging the US government to lift the ban, as well as export control experts who argue that the government lacks legal authority to impose it. The outcome from this meeting could help determine how aggressively governments regulate the next generation of frontier AI systems.
Behind the ban
When Anthropic released Fable 5, the company framed the model as a safer version of the Mythos-class models, noting that it has guardrails designed to ensure public safety. On top of that, the model was initially released for a limited time.
However, concerns soon emerged from Amazon's cybersecurity research team, citing concerns that the model could be jailbroken. That prompted Amazon’s CEO to escalate the issue to White House officials, leading the government to order that the model be restricted to all foreign nationals.
Officials argued that if foreign adversaries gain access to such capabilities, they could pose risks to national security. And while Anthropic complied with the ban, including taking Fable offline for everyone while it works on filtering non-US nationals in real time, the company isn’t happy with the decision.
Anthropic’s race to stay afloat
The case with Fable 5 isn’t the first time the US government has made such a consequential decision for Anthropic. In March, the Pentagon designated the company a supply chain risk, mandating that all federal agencies stop using Anthropic’s AI models after the company declined to let its models be used for autonomous weapons or mass domestic surveillance.
And just as Anthropic refused to let that issue go unaddressed, the company has now met with White House officials to find common ground on this new development. Details of the meeting have yet to be made public, but Reuters says that the government is seeking assurance from Anthropic that the model won’t be used to harm US citizens.
The publication also says that Commerce Secretary Howard Lutnick was part of the process and is expected to meet with Dario Amodei during the G7 meetings in France, where negotiations may continue. Also at the meeting was National Cyber Director Sean Cairncross.
Support grows for Anthropic
Anthropic's pushback to the restrictions has attracted support from several cybersecurity leaders, who signed a letter urging the US government to reverse the measures. These security leaders backed Anthropic, noting that the company “built multiple protections into the Fable model to prevent its use for cyber offensive uses.”
Corroborating Anthropic’s stance on the ban, the group contends that the jailbreak cited by White House officials also applies to other models. The letter also warns that removing a model as good as Fable in cybersecurity tasks while the US “adversaries are rapidly advancing is dangerous.”
Anthropic has also received backing from export control experts questioning whether the government has the legal authority to impose the restrictions in their current form. Their objections add another layer to the dispute, which now spans technical, legal, and national security concerns.
As negotiations continue, the case is emerging as a defining moment in the growing tension between AI innovation, national security, and regulatory authority. Whatever becomes of this will play a pivotal role in the future of AI access.
Also read: Anthropic’s confidential IPO filing puts the Claude maker’s valuation, revenue growth, and public-market timing under closer scrutiny.