Microsoft president Brad Smith stated that consumer products like milk have more oversight than AI companies and argued that the U.K. is best placed to lead global safety rules. Speaking on a BBC podcast, Smith called artificial intelligence both a weapon and a tool putting cybersecurity at risk, asserting that the moment has arrived to institute global safety rules. He highlighted Britain's AI Security Institute, launched in 2023, as a leader in safety testing that receives access to major frontier models before deployment.
Smith's endorsement follows recent geopolitical tensions and industry developments surrounding AI oversight. U.S. President Donald Trump previously warned against globalist organizations and reportedly asked Anthropic and OpenAI to withhold new models from British testers, though the AI Security Institute confirmed its continued access. Meanwhile, tech leaders including Anthropic's Dario Amodei and OpenAI's Sam Altman have called to slow development following incidents with rogue AI agents. Trump also hosted major tech executives at the White House to sign a voluntary agreement involving independent auditors.
Smith dismissed voluntary self-regulation because it only captures responsible companies, advocating instead for legal obligations like mandatory breach alerts, outside evaluations before product releases, and kill switches. He believes Washington lacks the unique geopolitical positioning of London to broker international standards without the heavy-handed perception associated with Brussels. Smith hopes Microsoft and the U.K. can help bring people together to establish formal safety measures.
Why it matters: Microsoft's push for formal U.K.-led oversight highlights growing industry divisions over whether voluntary self-regulation is suffici