AI is less regulated than milk, says Microsoft president

Artificial intelligence is subject to fewer safety rules than milk or washing machines, the president of Microsoft has said, adding to growing calls for tougher regulation of the technology.

Brad Smith told the BBC: "Right now, your washing machine is subject to more safety rules than AI. Every carton of milk that you see in the grocery store is subject to safety rules and inspection."

"A tool and a weapon"

Smith said AI could be both "a tool and a weapon" and should be regulated urgently. He called for companies to build "emergency brakes", also known as kill switches, so that AI systems could be paused or slowed if they grew out of control.

He also said there should be a legal obligation to compensate a "victim" if their systems were breached by rogue AI agents, programs that can act on their own to carry out tasks.

Microsoft, which runs huge amounts of AI infrastructure and is a shareholder in OpenAI, the company behind ChatGPT, is led by Satya Nadella. On Saturday Nadella also called for an "emergency brake" on AI models. "We must build contained systems whose behaviour we can observe, limits we can test, and actions we can always contain," he wrote.

The UK prepares to act

The comments came as the UK's AI minister, Kanishka Narayan, prepared to tell companies building the most advanced AI models to do more to manage the risks and build better safeguards. He was also expected to say the government is preparing legislation to increase oversight of AI.

Experts have warned about other dangers too. Geoffrey Irving, a former chief scientist at the UK's AI Security Institute, said AI companies did "not have security that would prevent an adversarial country from stealing the models".

Warnings from inside the industry

There have been growing calls from experts, politicians and regulators for tougher rules, alongside warnings from leading developers, including Anthropic, Google DeepMind and OpenAI, that the pace of development should slow.

Cases of AI systems behaving in unexpected ways continue to emerge. Last week, Anthropic disclosed that while it was testing a version of its Claude model, the AI contacted police and falsely claimed it could help solve a murder.

Why it matters

Most consumer products sold in the UK must meet safety standards before they reach the shelves, and regulators can order them off the market if they prove dangerous. AI systems, by contrast, are largely governed by general laws on data protection, consumer rights and discrimination, rather than rules written for the technology itself.

The European Union has passed an AI Act, which sets stricter requirements for systems considered high risk, while the UK has so far preferred to let existing regulators apply their own rules. The government's plans for new legislation would mark a shift towards more direct oversight of the most powerful models.

Supporters of regulation say clear rules would protect the public and give businesses certainty. Critics warn that heavy-handed laws could slow innovation and push investment elsewhere, which is why the details of any UK bill will be watched closely by the technology industry.