Microsoft president says UK is best placed to drive AI regulation

Microsoft President Brad Smith has issued an urgent call for global artificial intelligence regulation, asserting that the United Kingdom is uniquely positioned to spearhead these critical safety rules. Speaking on a recent BBC podcast, Smith argued that consumer products like milk and washing machines currently face more stringent oversight than the burgeoning AI industry. He characterized AI as both a "weapon and a tool" capable of putting global cybersecurity at risk, emphasizing that the "moment has arrived" for establishing international AI safety protocols.

Smith's stark comparison highlights an alarming disparity in legal accountability for AI developers, contrasting it with well-established consumer product safety standards. He underscored the profound dangers inherent in advanced AI, warning of its dual potential for immense benefit and severe harm, particularly in the realm of global cybersecurity. The industry, he suggested, cannot rely solely on voluntary commitments to manage these escalating, existential risks.

Central to Smith's argument is the UK's distinct advantage in leading this regulatory charge. He pointed to Britain's AI Security Institute (AISI), established in 2023, which has rapidly emerged as a global leader in AI safety testing. Crucially, the AISI benefits from unparalleled access to cutting-edge models from top U.S. frontier labs, allowing it to evaluate these advanced systems for potential risks before they are publicly deployed. This early access is a cornerstone of its effectiveness.

This leading role, however, has recently faced political pressure. Reports indicated that U.S. President Donald Trump sought to prevent companies like Anthropic and OpenAI from sharing new models with British testers, decrying "globalist" organizations for allegedly interfering with American businesses. Despite this reported pushback and the potential for a diplomatic incident, the AISI confirmed on Monday that it continues to successfully access and test the latest AI models, underscoring its established partnerships and vital role.

Smith elaborated on why the UK's geopolitical standing makes it an ideal facilitator for AI regulation. He noted that Washington, D.C., perceives London differently from Brussels, suggesting this distinction offers the UK greater flexibility and potentially less political friction when pursuing international agreements. This unique position, he contends, allows Britain to foster a more globally cohesive and less encumbered approach to AI governance, circumventing some of the gridlock seen elsewhere.

The Microsoft president also addressed recent efforts by the U.S. administration to promote AI safety. Following a White House meeting last month, tech giants including OpenAI, Anthropic, Nvidia, AMD, SpaceX, Meta, and Google signed a "morally binding" document agreeing to collaborate with independent auditors for AI model assessment. While acknowledging this as a positive initial step, Smith firmly asserted that such voluntary self-regulation, by its very nature, is insufficient.

He cautioned that relying solely on good intentions meant "only the ‘good companies’ will comply," leaving less responsible actors unchecked and the public at risk. Smith, therefore, advocated for concrete, legally binding regulations to ensure comprehensive safety across the entire industry. This perspective aligns with growing calls from other prominent tech figures, such as Anthropic CEO Dario Amodei and OpenAI's Sam Altman, who have urged a more measured pace for AI development in light of past incidents involving "rogue AI agents" — a term referring to AI systems that act autonomously in unintended or harmful ways.

Smith outlined several specific safety measures he believes are essential for effective regulation. These include a legal mandate for organizations to immediately report if their AI systems are breached by rogue agents, a requirement for external evaluators to approve AI products before public release, and the implementation of "kill switches" to halt or slow AI operations if human control is compromised or lost. He urged a calm but firm, collaborative international effort, envisioning Microsoft and the UK playing pivotal roles in uniting diverse stakeholders for this critical endeavor.

FAQ

Q: Why does Microsoft President Brad Smith believe AI regulation is urgently needed?

A: Smith argues that essential consumer goods like milk and washing machines have more regulatory oversight than AI companies. He views AI as both a "weapon and a tool" that poses significant cybersecurity risks and stresses that voluntary safety agreements are insufficient, as less responsible companies may not comply.

Q: What makes the UK uniquely positioned to lead global AI regulation, according to Smith?

A: Smith points to the UK's AI Security Institute (AISI), which has become a leader in AI safety testing by gaining early access to frontier models from U.S. labs before deployment. He also suggests that the UK's political standing means it is not viewed by Washington D.C. in the same way as Brussels, potentially giving it more flexibility to forge international agreements without facing similar political friction.

Q: What specific AI safety rules does Brad Smith propose?

A: Smith supports several key regulations, including a legal obligation for organizations to report breaches by rogue AI agents, requiring external evaluators to approve AI products before their release, and mandating the inclusion of "kill switches" to pause or slow down AI systems if human control is lost or compromised.