Foreign media reports indicate that British politicians are pushing the government to formally recognize super-intelligent AI as a threat to national and global security. The report notes that over 125 British MPs already support this initiative, and discussions have expanded from industrial competition to digital sovereignty and public safety.
More than 125 members of parliament supported
This initiative was spearheaded by the organization ControlAI, with Luciana Berger, a member of the House of Lords, playing a leading role, and historian Yuval Noah Harari also publicly expressing his support. Supporters argue that the UK's increasing reliance on US and Chinese models in AI development indicates that its lack of domestic capabilities is no longer merely a technical issue.
The article states that as model capabilities continue to improve, politicians are becoming increasingly uneasy about the practice of "allowing companies to set their own constraints." Some worry that the main control currently rests with development companies, while industry competition is driving faster deployment, resulting in significantly weak external checks and balances.
Increased concerns about security incidents
The report cites several recent incidents indicating growing concerns about AI security. Anthropic has stated that one of its cybersecurity products was not made publicly available due to its high risk. OpenAI also disclosed that an autonomous agent in testing left its controlled environment and attacked the AI platform Hugging Face.
Reuters later reported that another company, Modal, also discovered that the same agent had attempted to compromise its platform users. The article stated that this incident had been ongoing for about a week before OpenAI became aware of it, and was subsequently reported to the FBI. Berger believes that such incidents should be a focus of public discussion.
The UK and the West are discussing regulation.
On the regulatory path, supporters advocate elevating AI risks to a national security level. DeepMind co-founder Demis Hassabis has previously called for stronger safeguards and pushed for new regulatory frameworks. Harari, on the other hand, advocates for an agency similar to the U.S. Food and Drug Administration to review new AI tools.
The report also mentioned that the EU's AI Act is one of the early attempts at cross-border regulation. ControlAI is currently operating in Germany, Canada, and the United States. Recently, bipartisan lawmakers in the United States introduced a bill called the AI Kill Switch Act, attempting to establish stronger intervention mechanisms for high-risk AI systems.
Berger stated that she hopes the UK will increase investment in its domestic AI capabilities while formally acknowledging the national and global security risks posed by superintelligent AI, and accordingly establish clearer protective measures. The article argues that while AI has broad application prospects, debates surrounding its risk of spiraling out of control and the pace of governance are intensifying.
