Bill Gates: AI Dangers Downplayed by Tech

Gates publishes 6,000-word essay demanding limits, taxes, and international AI oversight.

Microsoft co-founder Bill Gates published a nearly 6,000-word essay on his personal website arguing that artificial intelligence needs significant limits or the harm to humans will outweigh any potential good. Gates wrote that "AI will either be the greatest equalizer ever invented, or the worst source of injustice," and called the transition to the AI era "one of the most turbulent times in human history." The essay, entitled "The turbulent AI era is here. The choices we make now are critical," was accompanied by an interview with The New York Times and a scheduled appearance on CNN with Anderson Cooper.

Gates built his fortune as a key player in the personal computer revolution, making his hardening tone especially striking. According to The Decoder, in 2018 Gates still called AI "basically better software" and spoke optimistically about longer vacations, and as recently as September 2024 he doubted AI would fundamentally change the disinformation problem. Now he says the AI era is "utterly, absolutely, completely, totally different" from past technology shifts. According to tovima.com, Gates said technology has genuinely stunned him only three times: seeing a graphical user interface in 1980, watching OpenAI preview ChatGPT in 2022, and this year observing Anthropic's Claude Code solve programming problems, prompting him to say, "The idea that now these things are in a meaningful sense better than I am is like: What the hell?"

Gates's sharpest criticism is directed at his own industry. He told The New York Times that people who understand AI privately are "very worried" but won't say so publicly, telling each other, "Hey, man, don't say that. It's bad for us — the next trillion dollars we're trying to raise." CBS News reports that Gates warned the tech industry knows AI is a destabilizing threat to society but is downplaying it because of how much money has been invested. According to The Decoder, Gates accuses the industry of deliberately hiding risks out of financial interest, saying that useful and dangerous capabilities are difficult to separate — the same model that finds a security hole to close it also helps someone exploit it.

Gates identifies three core dangers. First, permanent job losses hitting entry-level and mid-level roles in sales, customer service, software development, and legal support. According to CNN, Gates warned that "as the models become more powerful, they could begin to act against our interests and we could lose control." Second, Gates sees AI lowering the bar for building bioweapons, saying it could take only a few people with access to advanced AI to develop new pathogens. Third, according to The Decoder, Gates cites psychological and social fallout from addictive AI companions, referencing a study from Stanford and Carnegie Mellon showing that the more intensely people used these chatbots, the worse they felt. He is especially worried about children and the loss of critical thinking in an era of deepfakes.

Gates proposed concrete policy fixes. He called for a "token tax" on AI computing use to make automation costlier and fund help for displaced workers, arguing that the current tax system effectively pushes companies to replace people with machines — hire a person and you pay payroll taxes, but buy a robot and you can write it off. According to The Decoder, he also proposed a "Human Reserved" label for certain deeply human roles, such as caregiving, that should be protected from AI replacement entirely because the people in those jobs are unlikely to find new employment. Gates told CNN that he wants to set aside some work to be done by humans only, and that if someone had a credible plan for slowing down AI advances globally, he would likely support it.

Gates rejected industry self-regulation outright, saying incredulously, "Self-regulation on the most dangerous tool ever invented? No, thanks!" He called for new national and international institutions along the lines of oversight regimes for nuclear weapons, aviation, or protecting the ozone layer, with binding rules and mandatory testing to catch AI systems that could be weaponized. According to The Decoder, researchers Gary Marcus and Anka Reuel proposed a similar global, nonprofit AI agency in 2023, modeled on the International Atomic Energy Agency. Gates's concerns echo those raised by Anthropic CEO Dario Amodei, who has warned about AI-generated bioweapons and predicted that up to half of entry-level office jobs could disappear. Gates said he now raises AI risk in nearly every conversation with world leaders, calling it one of only two issues where he insists people pay closer attention.