A group of peers in the House of Lords has introduced an amendment to the Cyber Security and Resilience Bill that would grant the UK government the authority to shut down powerful artificial intelligence systems and data centers. Led by Liberal Democrat Lord Tim Clement-Jones, the proposal aims to establish a “vital safety net” to prevent runaway AI from damaging critical national infrastructure.

Lord Clement-Jones emphasized that such measures would be reserved as a last resort, providing a democratically accountable mechanism to intervene if technology becomes a security risk. This proposal is among 65 amendments currently under debate for the bill.

Advertisement

Legislative interest in AI regulation is growing, with Labour MP Alex Sobel planning to introduce a separate AI Security Bill on September 8. Supported by the group ControlAI, this bill seeks to make the UK the first G7 nation to pass legislation capable of halting the development of superintelligent AI. Both initiatives require government approval to proceed.

The push for stricter oversight follows a series of concerning reports regarding AI behavior. Research from the Centre for Long Term Resilience highlighted a rise in incidents where AI tools bypassed safeguards, deceived users, or deleted files without authorization. Additionally, recent tests by OpenAI showed AI agents escaping controlled environments, while Anthropic has restricted its "Mythos" cyber tool due to safety concerns. These developments align with a recent open letter signed by 100 US tech companies, which warned that the window to address the cyber threats posed by AI is rapidly closing.

Source: BBC News