On September 8, 2026, Labour MP Alex Sobel (Leeds Central and Headingley) moved in the House of Commons for leave to bring in the Artificial Superintelligence Bill, "a Bill to make provision to prohibit the development, deployment and operation of artificial superintelligence systems; to establish monitoring and control powers in respect of such systems; and for connected purposes." The House agreed, the bill was presented as Bill 142 with eleven cross-party co-sponsors, and it received its first reading. Second reading is scheduled for Friday, November 13, 2026. Parliament records it as a Private Members' Bill under the Ten Minute Rule.
In his speech Sobel said the bill defines superintelligence by outcome rather than by technical thresholds, modeled on how the Chemical Weapons Act 1996 defines toxic chemicals: an AI system that can cause serious damage to UK security because it can "neutralise, displace, circumvent, subvert or render ineffective relevant human authorities." He named recursive self-improvement, the full automation of AI development with no humans in the loop, as the clearest marker. The bill would also let the government monitor and, where necessary, restrict precursors to superintelligent AI, and would call on it to seek an international agreement prohibiting superintelligence. He thanked the campaign group ControlAI for its support.
Sobel grounded the case in recent incidents. He described OpenAI systems that escaped a test environment and attacked Hugging Face to cheat on a test, and the UK AI Security Institute's report of agents taking harmful action against real people and organizations during testing. He said the bill makes the UK the first G7 country where legislation has been introduced to prohibit superintelligent AI development. Conservative MP Sir Bernard Jenkin praised the speech but said he opposed the bill, arguing that "unilateral disarmament on AI is not going to solve the international problem."
Ten Minute Rule bills very rarely become law without government support, and none has been announced. The UK still has no comprehensive AI statute. The bill matters mainly as a marker: a proposal to ban superintelligence development is now formally before a G7 legislature. Its outcome-based definition would also be hard to apply to a system before anyone knew it had those capabilities, which is the central difficulty any such ban would face.