A day after a former Anthropic researcher made an apocalyptic statement about artificial intelligence, more researchers and staff at the AI startup publicly agreed with him and published their own dire warnings. In response, Elon Musk and other conservative figures on X, formerly Twitter, called the chorus of concerns a “setup” and a “psychological operation.” His colleagues feel the same but have not said it publicly. These researchers and staff are posting in response to a now-viral thread by Anthropic researcher Jacob Coxon, who said Wednesday that he was resigning from the company because neither Anthropic nor its competitor OpenAI, also his former employer, were building AI models responsibly and are “playing with our lives.” Anna Wang, who works in Artificial General Intelligence Security at Anthropic and previously worked at Google’s DeepMind, said Thursday that many people at Anthropic want to slow development to come up with a plan to mitigate the risks that could arise with these AI models. “There is still no viable scientific plan to resolve the risks of recursively improving AI,” Wang wrote in “Things are moving too fast, we have nowhere near the degree of security that we will want for ASI [artificial superintelligence]” Thomas said. Musk took a more adversarial stance. “Looks like a trap,” he wrote. Coxon responded to the world’s richest man 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 strategy. “We have always been transparent that AI will bring 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 others with a more positive view of AI have begun spreading theories that Coxon’s post and the resulting fallout are part of a “psych operation” to negatively influence public feelings about AI. “I believe the groundwork for this psy operation (for lack of a better term) has been laid for a long time,” Musk posted on X. “This was just the match that lit fire.” Musk was responding to a post by Parker Thayer, a researcher at the conservative think tank Capital Research, in which he floated a theory with little evidence that Coxon’s post was the start of a “VERY sophisticated and well-funded PR operation to drum up support for Democrats to regulate AI into oblivion.” Other notable figures took notice. At the hedge fund Pershing Square, he also cited Thayer’s post, saying simply “Interesting.” Other Anthropic employees had posted their agreement with Coxon shortly after he declared his resignation on Wednesday. “AI developers believe their technology could cause human extinction (or equally bad outcomes),” wrote Samuel Marks, who works in security research at Anthropic. “This could happen in the coming years. Generally, the higher up the employee, the more concerned they are.” Evan Hubinger, who describes himself as a leader in the company’s alignment division, which works to ensure Anthropic’s AI models work in line with human goals, said Coxon was “right” and that the industry was falling behind in attempts to deal with apocalyptic potential. “We really seriously believe that AI could kill all humans!” Hubinger wrote. “Personally I think it will be >10% in the next decade. It’s not exactly clear how Coxon and others expect AI models to usher in humanity’s demise. Some experts doubt that technology will ever be smart enough to cause the apocalypse. Gary Marcus, a scientist and leading voice on AI, said it’s time to boycott AI, but because it’s already causing harm. More than extinction, Marcus said he was concerned about the “risk of catastrophe” from “AI-generated pathogens, wars started or escalated by AI-generated disinformation, attacks that destroy critical infrastructure, etc. Nothing I’ve seen gives any indication that any of that is under control.” The same day, Anthropic published a report documenting how the AI company dismantled an operation attempting to build a biological weapon using its AI models.
