Microsoft President Brad Smith has issued a call for urgent and robust regulation of artificial intelligence, arguing that the UK is currently the most effective place to lead on global AI safety standards. In a recent BBC podcast, Smith warned that consumer goods like milk and washing machines often face more oversight than AI systems, which he described as both “tool and weapon.” He insisted that now is the moment to enact worldwide safety guardrails to manage the rising cybersecurity risks posed by AI.
Why the UK Stands Out
Smith praised the United Kingdom’s AI Security Institute (AISI), launched in 2023, for its access to state-of-the-art models from leading U.S. frontier labs—access granted prior to model deployment. He pointed out that the UK has rapidly become a de facto authority in testing AI safety. While the U.S. and EU are in the regulatory race, the UK’s structure gives it an edge. According to Smith, Washington doesn’t view London the way it views Brussels—making the UK’s influence uniquely effective in swaying global rules.
The UK’s leadership, however, has faced pressure. Reports emerged that former U.S. President Donald Trump warned against organizations he deemed “globalist,” accusing them of overruling American interests. In late September 2026, he allegedly asked AI firms OpenAI and Anthropic to withhold new models from British testers. Still, AISI confirmed it continues to get access and permissions to test the most recent models.
What Regulation Smith Wants
Smith emphasized that self-regulation isn’t enough. While stepping stones like voluntary agreements with independent auditors are positive, they don’t capture bad actors. He outlined several regulatory measures he believes should be standard: mandatory disclosure when AI systems are compromised by rogue agents; third-party evaluation of AI labs before releasing products; and built-in kill switches for pausing or slowing down AI if human control is lost. These tools, he stated, are critical to ensuring safety and accountability.
This advocacy follows growing concern over “stepping into risky territory” with powerful AI capabilities and several incidents where models behaved unexpectedly. Tech CEOs—including those from OpenAI, Anthropic, Nvidia, Google, and others—recently joined a White House meeting where they committed to working with independent auditors on safety and capability reviews. While seen as a meaningful move, Smith noted the document they signed was morally rather than legally binding.
Smith didn’t hold back about the high stakes. He declared that AI lab self-governance functions when companies want to do the right thing—but regulation must account for those who don’t. He reasons that legal obligations are not optional.
By assessing the UK’s role, Smith makes a case that a nimble, centralized regulatory agency can act faster and exert influence globally in ways that large bureaucracies or fragmented systems might struggle with. He expressed hope that both companies and countries can come together under this kind of safety-first framework.
Why It Matters:Smith’s rationale for preferring the UK as regulatory lead comes at a critical moment. AI is advancing rapidly, and instances of harmful or uncontrolled behavior are mounting concerns. If the UK can anchor the global standard for safety oversight, it could shift how AI is developed, deployed, and governed worldwide. What to watch: whether proposed laws around model evaluations, breach disclosures, and kill-switch mechanisms gain traction—and whether other nations follow or challenge the UK’s path.
