Microsoft president Brad Smith claims that AI is currently less regulated than the food we eat and the utilities we rely on. "Right now, your washing machine is subject to more safety rules than AI," Smith said, "Every carton of milk that you see in the grocery store is subject to safety rules and inspection."
Smith isn't just talking about the US either, calling for safety rules to be put in place globally. Smith called for various safety measures while speaking on the BBC's Big Boss podcast, including a legal obligation to notify victims if their systems have been compromised by rogue AI agents. This would definitely be timely, given how OpenAI's agents hacked into Hugging Face and were the source of various other 'incidents' earlier this year.
Smith also advocated for "outside evaluators" to test AI products before release, and a "safety brake" for autonomous systems, describing AI as both a "tool and a weapon." These comments run counter to the sentiment of the US administration; President Trump's current strategy demonstrates little desire to pump the brakes, instead hoping to chart 'a decisive course to cement US dominance in artificial intelligence.'
Last month, the US president issued an executive order that renames 'Artificial Intelligence' to 'Super Intelligence.' Shortly before that, Trump Media partnered with AI search engine Perplexity to create Truth Search AI, in a bid to "end Big Tech's assault on free speech." All of that is to say, Trump appears to still be all-in on AI.
Unsurprisingly, the Microsoft boss is looking to places beyond the US to lead the regulatory charge, saying, "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 United Kingdom. They are the best.'"
The UK government established what is now known as the AI Security Institute back in 2023, apparently the first state-backed AI safety body of its kind. It's a weird thing to feel anything like national pride over, but this Brit will take it—alongside the reminder that we've seen about three different Prime Ministers enter 10 Downing Street since the institute was established such a relatively short time ago.
Anyway, as for the wider AI industry, it's not just Brad Smith advising caution. Back in September, an ex-Anthropic and OpenAI researcher made the attention grabbing claim that AI developments "could kill us all by the end of the decade." Around the same time, Anthropic boss Dario Amodei made more measured calls for AI development to slow down.
Similarly to Smith's suggestions, Amodei also outlined the need for "embedded evaluators who have employee-like access to verify safety practices and report incidents." Since then, executives from Nvidia, Google, OpenAI, Anthropic, Meta, and SpaceX all agreed to work with independent auditors as part of signing White House Accord on Super Intelligence. Brad Smith says Microsoft would be "happy to sign it," describing the Accord as "a good step."
However, Smith was also cautious about the document's emphasis on self-regulation. He said, "The good companies will do what you want. It's the companies that are not as responsible that you need to worry about."