UK government ministers and Members of Parliament are proposing emergency powers to shut down advanced AI systems and data centres [1].

The move represents a significant shift in how the state manages emerging technology by granting the government direct authority to disable private infrastructure during a crisis.

The proposal comes as an amendment to the Cyber Security and Resilience Bill [1]. If passed, the amendment would allow the government to intervene when AI systems are deemed a threat to national security, public safety, or critical infrastructure [1], [2].

Lawmakers said the rapid advancement of artificial intelligence could create risks that existing regulations cannot mitigate, particularly during major national emergencies [2]. The proposed powers would specifically target the physical and digital infrastructure that supports advanced AI models [1].

Under the proposed framework, the government would have the authority to disable data centres to stop the operation of rogue or dangerous systems [1]. This "kill switch" capability is designed to prevent AI from causing systemic failures in critical services, or compromising state security [2].

While the bill focuses on resilience, the inclusion of these emergency powers marks a more aggressive stance on AI governance [1]. The debate centers on the balance between fostering technological innovation and maintaining the ability to neutralize threats in real time [2].

Government officials said the measures are necessary to protect the public from unpredictable AI behavior during high-stress scenarios [2]. The specific triggers for such a shutdown remain a point of discussion within Parliament [1].

UK government ministers and Members of Parliament are proposing emergency powers to shut down advanced AI systems.

This proposal signals a move toward 'hard' regulation of AI, shifting from guidelines and safety standards to direct state control over the physical hardware. By targeting data centres, the UK government is acknowledging that software-level safeguards may be insufficient, necessitating a physical or network-level override to ensure national security.