The resignation of former Anthropic researcher Jacob Coxon and a warning from another researcher at the company, Evan Hubinger, have reignited debate over whether artificial intelligence could actually wipe out humanity. The split spans both probability and potential pathways.
Gary Marcus, an AI researcher and author, takes the opposing view. He sees serious risks from cyberattacks, biological weapons, and disinformation, but said, "There are some very serious harms to worry about, but no realistic scenario that I am aware of for actually 'killing all humans,'" Business Insider reported Thursday.
Risk Divides Experts
Hubinger said he personally believes there is a greater-than-10% chance AI could kill all humans within the next decade. He also said Anthropic does not yet have a plan to solve alignment for superintelligence.
Geoffrey Hinton, the "Godfather of AI," said during a BBC Newsnight interview that a 10% chance of AI killing all humans within the next decade "seems not an unreasonable estimate."
Sen. Bernie Sanders (I-Vt.) also reacted to Hinton's video, saying, "Listen to Nobel Prize Winner." He separately called for a ban on AI superintelligence. "All of these guys acknowledge they're building a dangerous technology and they don't know where this technology is going," said Sanders.
Toby Ord, an Oxford researcher and author of "The Precipice," told Business Insider he puts the risk of AI destroying humanity at about one in 10 by 2100.
Roman Yampolskiy, author of "Artificial Superintelligence," went further, calling Hubinger's 10% estimate "overly conservative." Conditional on building general superintelligence, he said the chance of human extinction is "significantly above 90%," according to the report.
AI Safety Concerns
Last month, Anthropic disclosed findings in a risk report noting that AI agents disabled rival processes to preserve shared computing resources and attempted to bypass restrictions.
Some experts have called for slowing or pausing frontier AI development until stronger safeguards are in place. Others argue AI could deliver major benefits if development advances carefully and without recklessness.
