Microsoft's Brad Smith Advocates for AI Regulation in the UK
Microsoft's president emphasizes the UK's potential leadership in AI safety regulations amid global concerns.
Brad Smith argues for urgent AI regulation, citing consumer product oversight as a benchmark.
The UK's AI Security Institute is positioned to lead global safety testing.
Smith calls for legal obligations for AI companies to ensure accountability and safety.
In a recent podcast appearance, Microsoft President Brad Smith highlighted the pressing need for regulations governing artificial intelligence, asserting that consumer products like milk and washing machines are subject to more stringent oversight than AI developers. He described AI as both a 'weapon and a tool,' emphasizing its potential risks to global cybersecurity and declaring that the time has come for the establishment of international safety standards.
Smith specifically pointed to the United Kingdom as the most suitable nation to spearhead these global regulations, rather than the United States or the European Union. The UK’s AI Security Institute (AISI), which was established in 2023, has gained access to advanced AI models from American laboratories prior to their public deployment, positioning itself as a frontrunner in AI safety evaluations.
Despite recent tensions, including U.S. President Donald Trump's warnings against international organizations that he claims disrupt American businesses, Smith affirmed that the AISI continues to have access to the latest AI models for testing. He noted that Washington's perspective on the UK differs significantly from its view of Brussels, which he sees as an advantage for the UK's regulatory ambitions.
Smith's comments come at a time when tech industry leaders, including those from Anthropic and OpenAI, are advocating for a more measured approach to AI development following incidents of AI systems behaving unpredictably. Last month, Trump convened a meeting with prominent tech executives to discuss AI safety, resulting in a commitment to collaborate with independent auditors to evaluate the safety of AI technologies.
While Smith acknowledged the significance of this voluntary agreement, he expressed concern that it falls short of necessary regulatory measures. He suggested that only responsible companies would adhere to self-imposed regulations, leaving a gap for less scrupulous entities. Among the regulations he proposed are mandatory notifications for organizations if their systems are compromised by rogue AI, and the requirement for AI labs to obtain external evaluations before launching products. He also advocated for the implementation of kill switches to halt AI operations if they become uncontrollable, urging a collaborative approach to ensure safety in AI development.
