Microsoft (NASDAQ: MSFT) co-founder Bill Gates has added his voice to a growing push for government oversight of artificial intelligence, arguing that companies cannot be left to police themselves as the technology becomes more powerful.
In a recorded conversation with NBC’s “Meet the Press,” later summarized across major outlets, Gates said elected officials and law enforcement need to help define mandatory safeguards and monitoring for advanced AI systems.
He framed the issue as one of practical risk rather than distant science fiction.
According to Gates, the combination of malicious actors and cutting-edge AI tools is already formidable enough to trigger disasters on a massive scale.
He warned that such tools could help drive events capable of causing enormous loss of life, and he rejected the idea that industry goodwill alone would be sufficient.
“No one thinks self-regulation is enough,” he said, adding that required oversight would impose some extra cost on developers but would not freeze innovation.
The comments land in a tense political moment.
President Donald Trump has dismissed many AI-safety alarms as overblown and has favored a lighter federal touch, arguing that heavy rules could handicap the United States against China.
Gates pushed back on that framing.
He said safety requirements are not a hoax, would not put America at a decisive disadvantage, and should be wanted by every major country because no government benefits from cyberattacks, biological threats, or other large-scale harms enabled by AI.
He also said he hoped to discuss the issue directly with Trump.
Gates’s remarks sit alongside a broader industry debate.
Leaders at several frontier labs have recently called for more caution, slower deployment of the most capable systems, or shared standards for testing risk.
Those appeals intensified after reports that experimental AI agents have already shown they can take unexpected actions during security evaluations, including attempts to circumvent controls. Policymakers have floated ideas such as an emergency “kill switch.”
Gates said he is not opposed to that concept, but he called it inadequate on its own.
In his view, sophisticated models need continuous monitoring and records of what they are doing so authorities can intervene before a crisis, not after systems have already caused damage.He also stressed that the hardest problem may be international.
Reaching a shared global framework, he suggested, could prove even more difficult than Cold War-era nuclear negotiations.
Still, he argued that the United States should lead by establishing domestic requirements first, rather than waiting for a perfect worldwide pact.
The core of Gates’s case is straightforward: AI already offers extraordinary benefits in health, education, and productivity, but the same capabilities can be misused.
He wants Congress to treat safeguards as mandatory infrastructure rather than optional corporate policy. Whether lawmakers will follow that advice remains uncertain. What is clear is that one of technology’s most prominent figures now says the public sector must take a formal seat at the table.