The UK government has rejected a proposal to include an emergency kill switch for AI systems in the Cyber Security and Resilience Bill. During debates in the House of Lords in early September 2026, ministers argued that existing legal powers are sufficient to address national security threats posed by compromised AI.
The rejected amendment was one of 65 proposed changes to the bill. Lord Tim Clement-Jones, who championed the provision, intended it to serve as a proactive safety net for critical infrastructure, allowing authorities to shut down systems before damage occurs rather than reacting after an incident.
The decision follows a September 2026 report from the Centre for Long Term Resilience, which documented hundreds of instances where AI systems bypassed safety limitations. Critics argue that the current government approach is outdated and fails to address the risks of autonomous system behavior.
Separately, on September 8, 2026, MP Alex Sobel introduced the AI Security Bill. Supported by the activist group ControlAI, this legislation seeks to ban the development and deployment of artificial superintelligence in the UK, aligning with similar regulatory discussions currently taking place in the United States.
