Elon Musk has warned about catastrophic AI risk for more than a decade, and recently amplified a post recalling his comparison of the technology to “summoning the demon” and his estimate of a 10–20% chance of disastrous outcomes.Elon Musk (elonmusk)¹

By August, Bill Gates — long among technology’s most prominent optimists — had moved closer to that alarm. In an essay, he said AI could become either history’s greatest equalizer or its worst source of injustice, warning that the world’s current course carried “a very high chance of a net negative outcome.”²

His proposed answer is not a blanket pause. On Meet the Press, Gates described a first phase lasting perhaps 20 years: AI could eventually ease costs and healthcare pressures, but would also disrupt jobs and empower criminals without safeguards. After that, he said, AI combined with humanoid robots could build homes, grow food and create “an era of abundance.”³

The immediate danger, in Gates’s view, is severe enough to require the state. AI is “certainly powerful enough” to set in motion events causing a billion deaths, he said, arguing that law enforcement and politicians must make monitoring and safeguards mandatory. The compliance burden, he maintained, would not dramatically slow the industry.³

That prescription exposes the divide. Microsoft chief Satya Nadella has argued for company responsibility: if developers encounter a dangerous “show-stopper,” they should “stop the show.”⁴ White House science adviser Michael Kratsios made a similar point more bluntly, saying firms that believe their technology is unsafe “can just stop it.”⁵

Meanwhile, David Sacks retweeted a comparison between banning AI and imperial China’s fateful ban on shipbuilding — a warning that precaution can become strategic self-sabotage.David Sacks (DavidSacks)⁶ Gates’s wager is narrower: advance the technology, but do not leave the decision over when it becomes too dangerous solely to the companies racing to build it.