Microsoft says milk has more rules than AI⁠.

Microsoft's president, Brad Smith, has said milk and washing machines are more regulated than AI.¹ "Right now, your washing machine is subject to more safety rules than AI," he told the BBC.² He called the technology both a tool and a weapon that urgently needs to be regulated.¹

Smith wants global safety rules, and he was specific about what they should contain.¹ Firms would be legally obliged to write incident reports and to tell a victim if their system is breached by a rogue AI agent.¹ Companies would need emergency brakes, known in the industry as kill switches, to pause or slow AI that gets out of control.² He also suggested independent government evaluators should vet new AI products, and that rules should decide when outside evaluators sign off before a release.¹

He named the UK as the country best placed to drive those rules, because of its AI Security Institute, set up under Rishi Sunak as the AI Safety Institute.¹ He said Washington does not look at London the way it looks at Brussels, and suggested the US would accept UK standards more readily than EU ones.¹ That puts him at odds with President Trump, who has said the sector should regulate itself so American firms do not lose out to Chinese rivals.¹

Self-regulation is the part he called insufficient.¹ At the end of last month the leaders of several US AI companies signed a "morally binding" document at the White House, agreeing to work with independent auditors.¹ Smith said Microsoft was happy to sign it, but that the problem with any self-regulatory regime is that only the good companies do what is asked.¹ "It's the companies that are not as responsible that you need to worry about," he said.¹

The interview follows a similar call from Microsoft's chief executive, Satya Nadella, who on Saturday said AI models should have an emergency brake.² It also lands on the day the UK's AI minister, Kanishka Narayan, is due to tell frontier AI developers to do more on safety, with reports that the government is preparing to legislate for more oversight.²

Nothing changes in law today

This is a call for rules, not a rule. No British body has announced anything new on the back of it, and a minister's speech is not a law. It is also worth saying plainly that a company the size of Microsoft can carry the cost of inspection far more easily than the smaller firms whose tools you may be buying. When the biggest supplier in the room asks to be regulated, it is fair to ask who the rules would hurt most.

The useful part is the list itself. A way to stop an agent mid-task, a written record when something goes wrong, and a duty to tell the people affected are things you can ask any supplier about now, without waiting for Parliament. If the answer to how they would stop it is vague, that tells you something about the product.

Also today

• UK AI minister to say firms cannot police their own models

  Kanishka Narayan will call on frontier AI developers to do more on safety, say Britain cannot rely on them to police their own models, and announce stronger cyber defences.³

• Cisco puts AI agents into Webex meetings and calls

  Cisco has added agents that join Webex spaces, meetings and calls to handle multi-step tasks, alongside a Google Meet partnership for its devices and a new customer service platform.⁴

• Amazon in talks to buy chip software firm Decart

  Amazon is in talks to buy Decart, whose software lets developers switch between different AI chips, in a deal the Wall Street Journal says could be worth around $7bn.⁵

• Nvidia weighs buying or deepening its stake in Reflection AI

  Nvidia is in early talks to invest more in Reflection AI, an open-weight model maker last valued at $25bn, or to buy it, though the talks could still fall through.⁶