Bill Gates Says AI-Enabled Bioweapon Could Be Worse Than Covid, Faults White House

The Microsoft co-founder spoke at the Financial Times' Future of AI summit on November 4, according to accounts of his remarks. He named two risks he considers most urgent. The first is AI tools helping people design biological weapons. The second is attacks that exploit cybersecurity weaknesses across society.

The remarks matter because of who is making them and when. Gates is among the best-known backers of pandemic preparedness, and he made the case for government-led AI safeguards a day after US midterm elections that have unsettled AI and chip investors. In recent months he has moved from general warnings to direct criticism of US policy.

Gates said the capabilities of today's AI systems are striking, and that neither their benefits nor their potential for harm should be underestimated. He singled out openly available models, saying developers can strip out their safety safeguards and monitoring.

He said international efforts to manage these risks are stuck as long as the United States stays out. He also rejected the idea that AI companies can police themselves. When many companies are involved, he argued, only governments can solve the problem.

The Trump administration has generally rejected stronger federal oversight of AI developers in favour of voluntary arrangements with industry.

Gates set out several steps he wants governments to take. They should set standards for judging the safeguards and monitoring that companies build into their models. They should lead a global effort to fix cybersecurity vulnerabilities. They should expand third-party audits and tighten security requirements for new models. And they should build large-scale surveillance of air, wastewater and blood so that an AI-assisted outbreak could be detected early.

He linked the urgency to the pace of model releases. He said he expects OpenAI and Anthropic to release new models this year, with two major releases expected in 2027. Losing control of such systems could become a serious danger, he said.

Gates also praised Anthropic for its openness about the risks. The company's IPO filing warns that the technology could pose "catastrophic or existential risks to humanity".

His warnings on this subject have been growing more pointed. In his annual letter in January, Gates wrote that a non-government group using open-source AI tools to design a bioterrorism weapon was now a greater risk than a naturally caused pandemic. In September he said AI was powerful enough to drive events that could cause a billion deaths, and he called on Congress to legislate.

Gates also used the summit to argue that AI should be pointed at beneficial goals such as drug development and shorter hospital waiting lists. He has said repeatedly that safeguards need not dramatically slow the industry.