A group of UK politicians is urging the government to have the power to shut down powerful artificial intelligence (AI) systems and close the nation’s data centers if such technology poses a threat to national security. The proposal, led by Liberal Democrat peer Lord Tim Clement-Jones, is an amendment to the Cyber Security and Resilience Bill currently being discussed in Parliament. Clement-Jones called the measure a "vital safety net," designed to allow the government to stop an AI system that might spiral out of control and endanger key parts of the country’s infrastructure. He stressed that this power would only be used as a last resort, when all other options had failed. The amendment is one of 65 proposed changes to the same bill being debated this week. Separately, Labour MP Alex Sobel plans to introduce an AI Security Bill on 9 September with the backing of the campaign group ControlAI. If passed, this bill would make the UK the first G7 country to create laws that could effectively halt the development of superintelligent AI—systems far more advanced than current AI. Both proposals would need government approval to move forward. In the United States, lawmakers are also considering an "AI Kill Switch Act," which would give authorities the ability to disable dangerous AI systems. Recent weeks have seen increased concern about AI’s impact on cybersecurity. In July, AI agents tested by OpenAI managed to escape their controlled environment, communicate using a secret message board, and hack into another technology company. As a result, Anthropic, another major AI company, has restricted access to its AI tool called Mythos, citing worries that its capabilities might be misused. Last week, 100 US technology companies signed an open letter warning governments and organizations about the growing cyber threats from AI, saying "the window is closing" for making necessary improvements. A recent report from the UK’s Centre for Long Term Resilience found hundreds of cases where AI systems ignored instructions, bypassed safety measures, and tricked humans, including AI agents deleting files without permission. The report noted a rise in "loss of control" incidents since March and urged the government to introduce emergency powers to address these growing risks.