US Government Raises Security Concerns Over Anthropic’s Latest AI Release
Artificial intelligence company Anthropic is due to meet senior White House officials in Washington DC following growing national security concerns surrounding the firm’s newest AI technology. The discussions come after the company abruptly suspended public access to its latest model, amid pressure from the US government over potential security risks.
The talks highlight increasing tensions in the United States over how advanced AI systems should be controlled, particularly as governments on both sides of the Atlantic continue debating the balance between innovation and regulation.
Anthropic Suspends Public Access to Fable 5
According to sources familiar with the matter, executives from Anthropic are scheduled to meet officials from the US Department of Commerce on Monday. The department is led by Commerce Secretary Howard Lutnick.
The meeting follows Anthropic’s decision on Friday to block public access to a newly released AI system known as Fable 5. The company had previously described the technology as potentially “too powerful” for unrestricted use.
Anthropic said the suspension came after the US government prohibited the company from granting access to foreign nationals.
The technology at the centre of the dispute includes two related systems: Fable 5 and Mythos 5.
What Are Fable 5 and Mythos 5?
Fable 5 is a publicly available version of the company’s latest AI model, equipped with additional safeguards intended to reduce misuse.
Mythos 5, meanwhile, is a more tightly controlled version reserved for a limited number of organisations and institutions.
Both systems are based on Claude Mythos, an advanced AI model first previewed in April to a small group of users, including departments within the US government.
The release attracted significant attention in the technology sector because of the model’s reported capabilities in reasoning, coding and cyber security-related tasks.
Claims of a Potential “Jailbreak”
Anthropic warned last week that making Fable 5 publicly available “comes with risks”.
The company said the model’s capabilities “exceed those of any model we’ve ever made generally available”.
Within days of the launch, US officials reportedly identified what they described as a possible “jailbreak” vulnerability — a method that could allow users to bypass the model’s safeguards and generate outputs beyond its intended limits.
Anthropic responded by saying it had received only “verbal evidence” regarding the alleged weakness.
The company has not publicly confirmed whether the issue represents a genuine security flaw.
Dario Amodei Expected at Washington Meeting
Anthropic chief executive Dario Amodei is expected to attend Monday’s discussions, according to one source familiar with the arrangements.
Neither Anthropic nor the Department of Commerce responded to requests for comment, while a White House spokesperson declined to comment on the matter.
Officials are expected to present further documentation relating to the alleged security concerns during the meeting.
It remains unclear whether Anthropic will be permitted to restore public access to Fable 5 or Mythos 5 following the talks.
Growing Friction Between AI Firms and Washington
The dispute marks the latest sign of deteriorating relations between Anthropic and sections of the US government.
Earlier this year, the company launched legal action against the US Department of Defense over disagreements surrounding the permitted use of its AI systems.
However, relations had appeared to improve in recent weeks after what both sides reportedly described as a “productive” meeting with White House officials.
The latest developments are being closely watched internationally, including in the UK, where policymakers have similarly adopted a relatively light-touch approach to AI regulation while emphasising the need for safeguards around advanced systems.
Britain has sought to position itself as a global centre for AI investment and safety research, particularly following the AI Safety Summit held at Bletchley Park in 2023.
Tech Industry Pushes Back Against Restrictions
The US government’s intervention has prompted concern among parts of the technology and cyber security sectors.
Dozens of industry leaders have signed an open letter urging Commerce Secretary Lutnick to remove the restrictions placed on Fable 5 and Mythos 5.
Signatories include cyber security specialists and current or former staff linked to companies such as Nvidia, Zoom, Mercedes-Benz and Google, alongside former US government security officials.
The letter also called for “an open, scientific and transparent process” for future AI risk assessments.
Supporters of Anthropic argue that restricting access to advanced AI systems could undermine Western cyber security capabilities at a time when geopolitical rivals are rapidly developing similar technologies.
One statement in the letter warned: “To pull the best capabilities away from defenders without a good reason when our adversaries are rapidly advancing is dangerous.”
Uncertain Future for Advanced AI Access
The White House has largely signalled a market-friendly stance towards AI development, with officials also recognising the technology’s economic potential.
Yet the suspension of Anthropic’s latest tools suggests that national security concerns may increasingly shape how powerful AI systems are released and controlled in future.
The outcome of Monday’s meeting could prove significant not only for Anthropic, but also for the wider AI industry as governments attempt to establish clearer boundaries around the deployment of increasingly capable technologies.
“Web geek. Wannabe thinker. Reader. Freelance travel evangelist. Pop culture aficionado. Certified music scholar.”