Government officials within the Labour party have identified artificial intelligence as an emerging priority concern alongside economic pressures and cost-of-living challenges. Multiple cabinet ministers have privately flagged anxieties about concentrated AI power potentially reaching malicious actors, while others worry the administration is lagging behind public awareness of AI safety risks. The shift in focus intensified following statements from AI Minister Kanishka Narayan and MI5 Chief Ken McCallum earlier this week, with warnings that rogue agents could exploit AI for cyberattacks and that extremist groups might weaponize the technology to create bioweapons.
Political figures have attributed the renewed emphasis on AI safety to the recent change in party leadership. Liam Byrne, chair of the business, innovation, science and trade committee, noted that the new government under Andy Burnham is taking these warnings more seriously than the previous administration, which he characterized as overly aligned with major technology companies. This week, Byrne’s committee convened a parliamentary hearing with executives from Anthropic, OpenAI, DeepMind, and Meta, alongside senior civil service officials responsible for AI security.
A key tension has emerged regarding access to new AI models for government testing. The committee discussed whether such testing should be mandatory, and concerns were raised about the Trump administration’s reported requests to limit model sharing with the UK’s AI Security Institute. Despite these pressures, officials confirmed that cooperation with American companies has been maintained. Byrne emphasized that the government faces a strategic dilemma balancing continued access to cutting-edge models against implementing necessary safety measures.
Experts and officials have highlighted two distinct threat vectors. Geoffrey Irving, former chief scientist at the AI Security Institute, warned that open-source AI models—which are freely available and can be modified—are catching up to commercial counterparts and may already possess capabilities for serious harm. Meanwhile, bioweapon risks represent a potential next frontier as AI models grow more powerful, though experts note significant technical gaps between designing dangerous pathogens and deploying them as weapons.
The UK’s upcoming G20 presidency is being positioned as a critical opportunity to advance international AI regulation. Officials argue that momentum for action may shift if US congressional elections alter the political landscape, presenting a window for stronger governance frameworks on AI development and deployment.
Article Attribution | Read More at Article Source
Article summary produced by Claude AI