The US government has delivered a direct blow to the world of advanced artificial intelligence, forcing Anthropic – one of its leading developers – to disable two of its most sophisticated AI models for users globally. The sudden move has sparked concerns about access restrictions and control over cutting-edge tech. But what does this mean for everyday Brits, and how far-reaching are the implications?
Anthropic's decision follows a mysterious order from Washington, citing unspecified national security risks linked to its Fable 5 and Mythos 5 models. While the exact details remain unclear, it is believed that US authorities suspect these AI systems could be exploited – or 'jailbroken' – to uncover vulnerabilities in software.
This move marks a significant shift for the US government, which has historically regulated access to advanced technologies through export controls on physical components like chips. Now, they're targeting the AI models themselves, raising questions about who gets to control access to these powerful tools. The timing is particularly sensitive given Anthropic's recent stand-offs with parts of the US military over the use of its AI for surveillance and autonomous weapons.
Just days ago, Anthropic launched Fable 5 – a major upgrade to its 'Mythos-class' capabilities. This new tier had been touted as one of the most significant developments in AI history, but experts warn that its potential misuse could have devastating consequences. If misused, Mythos models like these could amplify sophisticated cyber-attacks on banking systems and other critical infrastructure.
Anthropic has responded with frustration to the government's directive, arguing that a 'narrow potential jailbreak' doesn't warrant recalling models used by hundreds of millions worldwide. They've also pointed out that rival AI providers offer similar capabilities – suggesting there might be more at play here than just technical concerns.