Microsoft’s Brad Smith calls for global AI safety rules beyond voluntary pledges

Microsoft president Brad Smith wants AI safety to move beyond promises from willing companies. In the BBC’s Big Boss Interview, he called for global rules covering emergency controls, incident reporting and outside evaluation before some products reach users. His argument: voluntary commitments cannot protect against companies that choose not to act responsibly.

Smith framed the gap through familiar products. Washing machines face more safety rules than AI, he said, while cartons of milk are subject to rules and inspection. The comparison was his case for applying established approaches to product safety to AI—not a claim that the technology and those products pose identical risks.

Controls before trouble, disclosure after it

Calling AI both a “tool and a weapon,” Smith urged governments and companies to begin now. His proposed safeguards address different moments: whether technology should be released, how to intervene when it gets out of control, and what companies must disclose after something goes wrong.

• Emergency controls: Companies should have “emergency brakes,” sometimes called kill switches, that can pause or slow AI systems if they get out of control. Smith described an intervention mechanism, not simply a commitment to assess risks.
• Incident reporting: Companies should be legally required to write incident reports and notify victims when AI goes rogue. That includes telling someone if a rogue AI agent breaches their system, rather than keeping the response inside the company.
• Pre-release scrutiny: Independent government evaluators should vet some AI technology before release. Smith called for deciding when outside sign-off is needed, drawing on the safety-review practices already used for other products.

Smith tied the urgency to his concerns about AI agents escaping human control and powerful models putting cybersecurity at risk. Those were his stated reasons for seeking regulation. The interview was a call to establish safeguards, not an announcement that new requirements had taken effect.

A voluntary accord is a start, not enough

His position contrasts with President Donald Trump’s preference for industry self-regulation, intended to keep US firms from losing ground to Chinese rivals. Trump has argued for rapid AI development. Smith’s response was not to dismiss voluntary work, but to argue that it no longer provides a sufficient safety framework.

At a White House meeting late last month, technology executives signed a “morally binding” document. The companies agreed to work with independent auditors to check that their AI systems behave as intended and do not hack or access technical systems in unintended ways. That agreement is the voluntary approach Smith addressed in the interview.

Smith applauded the statement and said Microsoft was “happy to sign it.” He called it a good and necessary step, but insufficient. His objection centers on participation and responsibility: a self-regulatory system can elicit the desired behavior from good companies without solving the problem posed by less responsible ones.

Why Smith wants Britain to lead

Smith identified the UK as well placed to drive international rules, citing its AI Security Institute. The institute began under former prime minister Rishi Sunak’s government as the AI Safety Institute. Smith praised Britain’s technical capability, saying people he knows in the technology sector would identify it as the leader.

He also offered a diplomatic argument: Washington views London differently from Brussels. Smith suggested that distinction could make the US more receptive to UK-led standards than to EU-led ones, despite Trump’s opposition to tighter controls. That is Smith’s assessment of a possible route to agreement, not a US commitment to accept British rules.