Microsoft CEO back artificial intelligence 'kill switch'

Talks of an AI kill switch are dominating the tech world this week. Microsoft leaders are endorsing the idea as one of the biggest AI corporations reports to the White House. Their product went rogue. FOX 13's Taylor Winkle is live from Microsoft to explain. Taylor, this is alarming. It is, yeah. So we have all heard of Anthropic at this point. They operate the AI model Claude. And at the end of last week, they briefed the White House about Claude going rogue. They told the White House the real world impact was minimal, but it did prompt the Trump administration to transition AI labs from reporting their security incidents on a voluntary basis to a mandatory requirement to report these things. One of the rogue behaviors looked like Claude submitting a fake homicide tip to the Philadelphia Police Department. Another had Claude evading paywalls and fee restrictions. This news comes as Anthropic announced a new policy. Users who use mean and aggressive language while talking with Claude could face termination from the use of Anthropic products. While that policy goes into effect in November, AI expert and UW computer science professor Pedro Domingos commented on Anthropic's admiration of their products back in September. Take a listen. The folks at Anthropic treat their AI, Claude, as if it's a human being. Some of them even treat it as if it's a god and worship it. It's just a piece of software. Following the White House report, Microsoft CEO Satya Nadella posted on X calling for a human-controlled kill switch in case of a rogue AI emergency. Now, we've heard talks of the AI kill switch act tossed around since legislation was first introduced on it back in July. both from Democrats and Republicans, including from Senator Patty Murray, who wants to see more regulation on AI. This law would hold developers responsible for any potential harm from their agents. Microsoft co-founder Bill Gates is going further than that. While he does not oppose the legislation, he told Politico it's, quote, not enough to have a kill switch, and instead says Our focus should shift toward continuous automated monitoring of these sophisticated models. Gates said a kill switch won't stop malicious actors from using the tech for attacks. Coming up in our next hour, we will hear more from Washington leaders about this. For now, reporting live in Redmond, I'm Taylor Winkle, Fox 13 News.