The Government who cried Fable: How the US/Anthropic saga could be felt worldwide

After Anthropic released its Fable generative AI model, the US government classified it as dangerous and told Anthropic to suspend use to all foreign nationals.

The AI company admitted the technology was extremely powerful and said the US authorities had raised security concerns just days after the model was publicly released. Anthropic met with senior White House officials on Monday over national security concerns about the new release.

According to WIRED, the government is requesting that all jailbreaks be blocked if Anthropic wants to re-release Fable 5 – something that may not be possible.

A growing interest in regulation

Fable is the constrained version of Mythos, the AI model first announced in April by Anthropic. Mythos AI was only released to a handful of organisations, with the company claiming it was too good at finding and exploiting cybersecurity vulnerabilities and so would be too dangerous in public hands.

Mythos experienced mixed reviews, with some analysts arguing over how dangerous the model could be in the wrong hands.

The suspension of Fable 5 and Mythos 5 models perhaps highlights how quickly security and regulatory concerns can impact even the most advanced AI systems. Despite Anthropic publicly disagreeing with the US government’s decision, access to the AI models remains suspended.

Sachin Puri, CEO at Network Solutions, suggested that – as these AI systems become more powerful – the conversation shouldn’t stop at capability, but should include governance, accountability and contingency plans for if guardrails fail.

“As AI lowers the barriers to both defence and attack, the industry has a responsibility to ensure that trust, security and protection become easier to access, not harder,” he said. “The future of AI cybersecurity will be defined not just by intelligence, but also by the governance frameworks, safeguards and trust infrastructure that surround it.”

The situation is being watched worldwide and could lead to further discussions about AI safety among researchers, regulators and large enterprises.

Global impact

The Anthropic news came shortly after the US government signed an AI executive order that made it mandatory for AI companies to licence their advanced models. US President Donald Trump’s administration continues to evolve its approach to AI regulation, leaving the rest of the world looking to see what comes next.

Anthropic has already become entangled in a lawsuit with the Pentagon. This has led to Europe watching the US AI market closely, as a region gaining traction quickly for its technology capabilities. From a UK perspective, industry leaders are looking at whether the right foundations are in place to enable the UK economy to retain meaningful capability.

“Sovereignty does not require doing everything alone, but it does require maintaining influence over the critical technologies and infrastructure that underpin economic growth,” explained Matt Harris, UKIMEA managing director of Hewlett Packard Enterprise (HPE)

The UK government has already taken significant steps to power up AI, including addressing planning barriers for AI data centres, establishing AI Growth Zones and creating specialist teams for infrastructure projects. In Harris’ mind, there is now a strong case for moving faster and being more ambitious.

“Policy intent must be matched by tangible outcomes,” he added. “The UK needs a clear view of which layers of the AI stack it wants to own, control or influence. It will be equally important to consider where else in the AI ecosystem the UK can build lasting strategic advantages in the years ahead.”

Anthropic will not be the only one embroiled in the AI debate. As global demand for these services soars, it’s inevitable that other AI models will also fall under the magnifying glass, thereby making concerns and consequences over safety even more urgent.

How to keep a cool head as the AI market heats up

Anthropic launches cybersecurity partnership with Nvidia, Cisco and others