A day after former Anthropic researcher Jacob Coxon warned of a potentially catastrophic future for artificial intelligence, several current employees and researchers at the company voiced similar concerns in public.
Musk and other conservative commentators on X, formerly Twitter, responded by describing the warnings as a coordinated “setup” or “psyop”.
The employees fear that the systems Anthropic is developing could become sufficiently capable and hazardous to pose a risk of human extinction within the next decade. They say many colleagues share those concerns but have not expressed them publicly.
Their comments followed a viral X thread in which Coxon said he was leaving Anthropic because neither it nor OpenAI, where he also worked, was developing AI responsibly and was “gambling with our lives”.
Anna Wang, who works on Artificial General Intelligence Safety at Anthropic and previously worked at Google DeepMind, said Thursday that many employees want development slowed while the company develops a strategy for managing potential risks.
“There is not yet a viable scientific plan to solve risks from recursively self-improving AI,” Wang wrote on X.
Drake Thomas, another Anthropic employee, said he respected Coxon’s decision not to work on these models if he believes they could create civilization-scale risks.
“Things are moving way too fast, we don’t have anywhere near the degree of assurance we’ll want for ASI [artificial superintelligence],” Thomas said.
Musk took a more adversarial approach. “Seems like a setup,” he wrote.
Coxon replied to Musk with a selfie, writing, “I’m real and these are my real beliefs. You could ask your xAI researchers about me if you hadn’t fired them.”
In a statement to the Guardian, an Anthropic spokesperson defended the company’s approach. “We have always been transparent that AI will bring both enormous benefits and unprecedented risks. To address these risks, we continue to build models with some of the strongest safeguards in the industry,” the statement said.
Musk and other advocates with a more positive view of AI have advanced theories that Coxon’s post and the ensuing controversy are part of a “psyop” designed to turn the public against the technology.
“I think the groundwork for this psy op (for lack of a better term) has been prepared for a long time,” Musk posted on X. “This was just the match that lit the fire.”
Musk was responding to a post by Parker Thayer of the conservative think tank Capital Research, who suggested with little evidence that Coxon’s post was the beginning of a “VERY sophisticated and well-funded PR operation to get support for Democrats to regulate AI into oblivion”. Other prominent figures also took note.
Bill Ackman, the billionaire chief executive of the hedge fund Pershing Square, also reposted Thayer’s post with the single word “Interesting”.
Other Anthropic employees voiced their agreement with Coxon soon after he announced his resignation on Wednesday.
“AI developers believe their technology could cause human extinction (or similarly bad outcomes),” wrote Samuel Marks, who works on safety research at Anthropic. “This could happen in the next few years. In general, the more senior the employee, the more concerned they are.”
Evan Hubinger, who leads Anthropic’s alignment division, which works to ensure that its AI models operate in accordance with human goals, said Coxon was “correct” and that the industry was failing to keep pace with the technology’s potential for catastrophic harm.
“We really do earnestly believe AI could kill all humans!” Hubinger wrote. “I personally think it is >10% within the next decade.
The precise mechanism by which Coxon and others believe AI could threaten humanity remains unclear, and some experts doubt that the technology will become intelligent enough to cause an apocalypse. Gary Marcus, a scientist and prominent AI critic, argues that the technology should be boycotted because it is already causing harm.
Marcus said extinction was not his primary concern. He is more worried about catastrophic risks involving AI-generated pathogens, wars triggered or intensified by synthetic disinformation, and cyberattacks that cripple critical infrastructure. “Nothing I have seen gives any indication that any of that is under control,” he said.
On the same day, Anthropic released a report detailing how it dismantled an operation that had attempted to use its AI models to build a biological weapon.