UK Removes ‘Safety’ from AI Body, Renames it AI Security Institute, Signs MOU with Anthropicl

14.02.2025

The U.K. government is making a strategic shift to prioritize economic growth and industry development through AI. As part of this transition, the Department of Science, Industry, and Technology has announced that the AI Safety Institute will now be known as the AI Security Institute. While the URL remains unchanged, the institute’s focus is shifting from existential risks and bias in AI to cybersecurity, specifically strengthening protections against threats to national security and crime.

Alongside this change, the government has entered into a new partnership with Anthropic. While no specific services have been confirmed, the memorandum of understanding outlines plans to explore the use of Anthropic’s AI assistant, Claude, in public services. Additionally, Anthropic will contribute to scientific research and economic modeling while also providing tools to assess AI capabilities in identifying security risks.

This shift aligns with the Labour government’s AI-driven Plan for Change, which was introduced in January. The plan notably omitted terms like “safety,” “harm,” and “existential risk,” signaling a broader move toward fostering AI-driven economic growth. The government aims to collaborate more closely with major tech companies while also nurturing its own domestic AI sector. As part of this initiative, civil servants will gain access to an AI assistant named “Humphrey,” and digital tools such as government-issued digital wallets and AI-powered chatbots will be introduced for public use.

Despite concerns that AI safety issues are being sidelined, officials maintain that the core mission of the AI Security Institute remains unchanged. Technology Secretary Peter Kyle emphasized that the institute will continue its work while ensuring AI is leveraged for economic expansion. Chair Ian Hogarth reaffirmed that security has always been a priority, highlighting the establishment of a new criminal misuse team and deeper collaboration with national security agencies as the next step in mitigating AI-related threats.

Meanwhile, perspectives on AI safety are shifting beyond the U.K. as well. In the U.S., the future of the AI Safety Institute remains uncertain, with Vice President J.D. Vance recently suggesting that its dismantling is under consideration. This reflects a broader global trend where governments are increasingly prioritizing AI’s role in economic and strategic development over existential risk mitigation.