Microsoft president Brad Smith has called for urgent global regulation of artificial intelligence (AI). In an interview published on Monday, 12 October 2026, on the BBC's Big Boss Interview podcast, he stressed the need for a global set of safety rules under which independent evaluators could test new AI developments. According to him, AI currently operates under even fewer requirements than those applied to simple products like milk. Reuters reported this. Smith is the president of Microsoft; his interview aired on Monday as part of the BBC's Big Boss Interview series.
Independent evaluation — the key proposal
One of Smith's key proposals in the interview was independent evaluation: new AI developments should be testable by independent evaluators. This means safety assessment would no longer remain an internal process of the company that created the product — it would be carried out by outside experts, independently of the company. Smith considers such an arrangement an integral part of a global set of rules. Independent audits, together with victim-notification requirements, a 'kill switch', and 'incident reports', are envisioned as forming a single global system.
The global rules Smith proposes
The global set of safety rules Smith deems necessary is built around three core mandatory requirements. The first concerns transparency: if a system is breached by a rogue AI agent, the victim — the party whose system was breached — must be officially notified. The second concerns a safety mechanism: every AI system must have a 'kill switch' — a mechanism that makes it possible to immediately stop a process that has escaped control or is acting dangerously. The third concerns reporting: companies would be legally required to file 'incident reports' on hazardous events that occur, meaning every serious incident must be documented.
In the interview, Smith described AI as simultaneously a 'tool' and a 'weapon'. He added that new AI developments should be testable by independent evaluators — safety assessment should be entrusted to outside, impartial experts rather than the developing company itself. Under such an arrangement, assessment results would be free of the developer's commercial interests and would become an integral part of the global set of rules.
The 'less regulated than milk' comparison
Smith backed his call with a stark comparison: AI, he said, currently operates under fewer requirements than those applied to simple products like milk. With this comparison, he drew attention to the gap between the seriousness of the technology and the level of its oversight. He warned that requirements placed on a technology described as both a 'tool' and a 'weapon' remain below those placed on everyday consumer goods.
"Microsoft president Brad Smith warns that artificial intelligence is currently regulated even less than the safety requirements applied to milk." — Reuters
Assessing the White House pledge
Smith also assessed the White House's 'morally binding' pledge on audits. He called the initiative a step forward but said he considers it 'insufficient'. Although Smith assessed the pledge positively, he does not see it as a complete solution. His position is that voluntary commitments alone are not enough — rules with legal force must be introduced at a global scale.
The UK's role
According to Smith, the United Kingdom is well positioned to take the lead in regulating AI. As the basis for this, he pointed to the country's AI Security Institute. In a statement to Reuters, UK AI minister Kanishka Narayan said that if the current voluntary system proves insufficient, the government will consider regulating advanced models. The minister's statement was published on the same day as Smith's interview — 12 October 2026. According to Narayan, the government will first observe the results of the voluntary system: if it proves insufficient, the question of regulating advanced models will be put on the agenda. Thus, on a single day, both the head of a tech giant and a representative of the British government stated their positions on AI regulation.
About the interview
Big Boss Interview is the BBC's series of conversations with the heads of major companies. The episode featuring Smith was published by the BBC on Monday, 12 October 2026; TechDefused also independently reported on the interview's content. In the podcast episode, Smith set out his views on AI safety in detail. Thus, Smith's call was covered on a single day in two independent sources — reports from Reuters and TechDefused.
