The Microsoft co-founder said that lawmakers must require safeguards and monitoring, arguing the burden on industry would be modest, as tech leaders, governors and a whistleblower clash over how fast AI should advance.
Microsoft co-founder Bill Gates has called on lawmakers in Washington to regulate the development of artificial intelligence, just weeks after leaders of top AI labs acknowledged the need to slow the pace of progress. In a recent interview, he said nobody believes self-regulation is sufficient.
Asked whether US Congress should pass legislation, Gates answered "absolutely." He said law enforcement and politicians must join the discussion on what safeguards and monitoring should look like, and that these should be mandatory. Such rules, he argued, would add some overhead for the industry without dramatically slowing its work.
Gates did not dismiss fears that AI could threaten humanity within years, but urged Americans to focus on a nearer danger: bad actors using the technology to cause mass harm today. He said AI is powerful enough to contribute to events killing a billion people, and that no weapon has matched the combination of malicious people and the latest AI tools.
Industry and political divisions
The remarks follow statements by the chiefs of Anthropic and OpenAI that the race to build powerful AI should slow. Those came after Jacob Coxon, a recently departed Anthropic whistleblower, said neither company was acting responsibly and that builders believe AI could kill everyone by the decade's end. Coxon later told "Meet the Press" he considers the lab leaders sincere in wanting to slow down.
House Speaker Mike Johnson, R-La., declined weeks ago to recall lawmakers to pass AI rules, saying he would defer to the companies. He said guardrails are needed, but that AI chiefs have not offered a federal legislative path and Congress should not rush into steps that could harm national security.
Anthropic CEO Dario Amodei has argued in an essay for regulation covering all US frontier AI companies, plus cooperation with China. OpenAI's Sam Altman told a United Nations gathering that international standards are needed for measuring capabilities, assessing risks and preserving human oversight. Meta's Mark Zuckerberg disagreed, saying industrywide coordination is unnecessary and each lab should pause internally when it spots problems.
States and Microsoft weigh in
Some states are moving ahead. California Gov. Gavin Newsom signed an executive order creating an expert board to plan stronger AI safety laws. Maryland's Wes Moore and New York's Kathy Hochul released their own plans this week; Moore's takes a three-pronged approach, while Hochul's creates an office to oversee AI companies' legal compliance.
At the UN, Microsoft President Brad Smith rejected the idea that companies must await federal rules before slowing down. Comparing the situation to two airplane makers racing while fearing a crash, he said the fundamental question is whether firms put public safety ahead of winning.
