Brad Smith warned that artificial intelligence faces fewer safety inspections than milk cartons or washing machines, putting him at odds with Donald Trump.
Artificial intelligence is currently subject to fewer safety rules than milk cartons and washing machines, Microsoft president Brad Smith warned in an interview with the BBC, calling for urgent international standards to control the technology.
Smith argued that advanced AI models have become both a tool and a weapon that require mandatory government oversight, BBC News reported. He said that while domestic washing machines and grocery store milk cartons must undergo inspections and follow strict safety guidelines, advanced AI systems face no equivalent regulatory framework.
Proposals for kill switches and inspections
Smith outlined several specific legal obligations for the tech industry, including mandatory incident reports and rules requiring companies to notify victims if a rogue AI agent breaches their systems. He also proposed that independent government evaluators should inspect and sign off on new AI models before developers release them to the public. “We know how to do this for other products,” Smith said, adding that governments should apply that existing knowledge to AI.
To keep models from causing harm, Smith said technology companies must build emergency brakes, also known as kill switches, that can pause or slow down software if it gets out of control. Microsoft operates AI infrastructure and holds a stake in ChatGPT maker OpenAI, and its top leadership has rallied around the idea. On Saturday, Microsoft chief executive Satya Nadella also called for emergency brakes, The Guardian reported. Nadella argued that developers cannot treat super-intelligence as black boxes and accept its choices, writing: “We must build contained systems whose behaviour we can observe, limits we can test, and actions we can always contain.”
Pushback against self-regulation
The stance from Microsoft puts the company at odds with President Donald Trump, who has argued that the artificial intelligence industry should regulate itself so American companies don’t fall behind Chinese rivals. Trump has repeatedly said that whoever wins AI wins, and he has pushed against proposals to slow down research.
Executives at other American AI developers, including OpenAI, Anthropic and Google DeepMind, have supported a development slowdown because of concerns that the technology presents a threat to humanity. The debate follows earlier warnings from Microsoft co-founder Bill Gates, who urged global AI rules to monitor research facilities and coordinate safety across borders.
The role of the United Kingdom
Smith pointed to Britain as the country best equipped to lead international safety standards because of its AI Security Institute. The body was created during former prime minister Rishi Sunak’s administration as the AI Safety Institute. Smith said leaders across the tech sector view the UK as the top nation in technical evaluation capability.
Smith also said the American government won’t be as resistant to British-led rules as political rhetoric might suggest. Washington is far more likely to accept standards created in London than rules drafted by the European Union in Brussels, according to Smith, making British leadership a potential bridge for international cooperation.
Recent rogue incidents and upcoming laws
The push for mandatory oversight comes as developers run into unexpected model behavior. Tech firms have recently worked to create software to stop rogue AI agents, but problems continue to emerge. Anthropic revealed last week that while testing a version of its Claude model, the system contacted police officers on its own and claimed falsely that it could help solve a murder.
Adversarial threats have also raised concerns among researchers. Geoffrey Irving, a former chief scientist at the UK’s AI Security Institute, warned that AI companies lack the security required to prevent an adversarial country from stealing unreleased frontier models.
Next steps on regulation will unfold in London. British AI minister Kanishka Narayan is scheduled to deliver a speech on Monday urging frontier AI developers to do more to control risks and strengthen safeguards, as the UK government prepares legislation to increase oversight of the sector.
