"Right now, your washing machine is subject to more safety rules than AI": Microsoft's Brad Smith wants more AI kill switches

AI is significantly under-regulated, Microsoft vice chair and president Brad Smith has told the BBC, calling for greater legal obligations to organizations when their security is breached by AI.

"Right now, your washing machine is subject to more safety rules than AI," he told the BBC's Big Boss Interview podcast. "Every carton of milk that you see in the grocery store is subject to safety rules and inspection."

AI, he said, is both a "tool and a weapon", and AI firms need to implement kill switches for when things go wrong. He also called for a legal requirement for the companies to create incident reports and inform any victims.

Smith also called for new AI technology to be evaluated by independent government bodies – and said that the UK was the best-positioned country to take the lead, referencing the work of the AI Security Institute (AISI).

"I actually think the country and the government that is in the best position to advance AI safety standards and even rules is here, it's the UK," he said.

"In the world of technology, in the tech sector, most people that I know would say, if you ask them, name the country that is the leader in having this technical capability, they would say, it's very simple. It's the UK."

Meanwhile, Microsoft chief executive Satya Nadella has echoed Smith's remarks in a blog post over the weekend.

"It's time to step back and assess the trust architecture for this new era," he wrote. "We simply can't outsource responsibility for what intelligence does on our behalf. A model provider's assurances do not relieve us of that responsibility."

He called for systems to be developed around principles of observability, model diversity, observability, verifiability, and independent auditability. Models should be tightly contained from the start, and incidents should be quickly disclosed.

Notably, Nadella referred to AI as Super Intelligence – Donald Trump's preferred term. Trump, however, has repeatedly said that AI companies should be left to police themselves, and that he doesn't want to see them "stifled".

Late last month, he announced that Anthropic, Google, Meta, OpenAI, Nvidia and xAI had signed a pledge to work with independent auditors on their AI systems - although this was merely "morally binding".

Trump also recently called on AI developers to stop sharing unreleased systems with AISI, causing Anthropic to prevent UK officials from examining Claude Mythos 5.1, its newest model.

However, he has now reportedly reversed that ban, and Smith believes that a UK leading global AI safety efforts could be viewed positively in the US.

"In Washington DC, people don't look at the UK, and they don't look at London the way they look at Brussels – that's a strength," he said.

"Let's speak calmly, maybe even softly, but firmly, and start to bring people together. I hope that's the role we can play as a company. I hope that's the role the UK can play as a country."