Microsoft’s president, Brad Smith, told the BBC that artificial intelligence currently operates under far weaker safety controls than items such as milk cartons or household appliances. He highlighted a stark contrast with President Donald Trump, who has argued that the United States should rely on industry-led self-regulation to avoid ceding advantage to Chinese competitors. Smith’s remarks place him at odds with the White House stance on AI governance.
Smith outlined a suite of measures he believes should become mandatory for AI developers. These include a legal duty to inform any affected party when a system is compromised by a rogue agent, the requirement to produce formal incident reports, and the installation of emergency “kill switches” that can halt a model’s operation. He also advocated for independent government evaluators to approve AI products before they reach the market, mirroring procedures used for other consumer goods.
According to Smith, the United Kingdom is uniquely positioned to spearhead a worldwide regulatory framework. He referenced the AI Security Institute, launched during former prime minister Rishi Sunak’s administration, as a foundation for building international standards. Smith praised the UK’s technical expertise and existing institutional structures, suggesting they give the country a natural advantage in shaping future AI safety policies.
While acknowledging the United States’ historical reluctance to adopt foreign standards, Smith argued that Washington might be more receptive to UK-led rules than to those originating from the European Union. He noted that American policymakers tend to view London more favorably than Brussels, a perception he believes could smooth transatlantic cooperation on AI oversight.
Smith’s call for stricter regulation aligns with a growing chorus of industry leaders, including CEOs from Anthropic, OpenAI and Google DeepMind, who have publicly supported a slowdown in AI development. In contrast, Trump recently convened executives from major U.S. tech firms at the White House, where they signed a “morally binding” pledge to work with independent auditors. Smith described that agreement as a positive but insufficient step toward comprehensive safety.
Concluding his interview, Smith warned that the current voluntary measures fall short of what is needed to protect global cybersecurity. He urged stakeholders to engage calmly yet decisively, emphasizing that coordinated international action is essential to prevent AI systems from operating beyond human control.