Second Anthropic Safety Researcher Resigns as Amodei Tells CBS Government Must Step In, OpenAI Confirms IPO Delay

Since Jacob Coxon's resignation from Anthropic on September 8 drew more than 100 million views on X, the fallout has widened. A second Anthropic safety researcher has quit citing the same concerns, CEO Dario Amodei has moved from an industry slowdown pitch to a direct call for government regulation, and OpenAI has said it won't launch its IPO this year.

A second resignation

Joe Benton, who worked on Anthropic's safety team, announced his resignation on Friday, September 11, according to the Associated Press via PBS. "Many of the people I know who work on safety research at AI companies want to do what is right for the world," Benton wrote. "But they feel their companies are trapped in a race to build superintelligence: either they stop and other, less conscientious people take their place; or, they continue, and risk participating in enormous harm themselves."

That framing echoes Coxon's, who wrote that both Anthropic and OpenAI "are racing straight to self-improving superintelligence and gambling with our lives," per PBS. Two departures in a week from the same company, both citing the same fear, suggests a pattern beyond a single resignation.

Amodei goes to CBS

Amodei's September 12 essay, which called for a three-step slowdown framework including independent evaluators with employee-like access, is now being followed by a direct television appeal. In an interview set to air on "CBS Sunday Morning" today, September 13, Amodei says "there needs to be involvement of the government" in AI safety, according to clips distributed by Whatfinger News and Social News XYZ.

That marks a shift from asking labs to self-police toward asking Washington to write rules. Amodei's essay already named international cooperation, including with "authoritarian" governments, as one of his three pillars, according to PBS.

The IPO backdrop

Anthropic is moving toward what could be the largest IPO in market history, potentially seeking a valuation near $2 trillion, according to Morningstar, which cites reporting from MarketWatch. Greg Martin, a managing director at Rainmaker Securities, told MarketWatch that Coxon's post "could force Anthropic to be more explicit about security in its IPO filing." Martin said the company will likely need to "beef up how they talk about safety and preventing some of the issues that are described in some of these tweets."

Meanwhile OpenAI's Sam Altman told Fortune his company will wait until next year to begin selling stock to investors, framing the delay around safety priorities, according to PBS. Altman also posted on X that he agreed with Amodei's call to slow capability development and to bring in independent evaluators with access similar to employees, saying more details would follow.

The Skynet framing

Anthony Aguirre, president of the Future of Life Institute, which called for a six-month AI development pause back in 2023, told PBS that the wave of resignations has forced a reckoning. "They've kind of realized, their employees have realized, everyone has realized that they're building Skynet," Aguirre said. "And in winning the race to Skynet, nobody wins. Really, nobody."

Evan Hubinger, Anthropic's Alignment Science lead, has publicly backed the underlying alarm rather than dismiss it. He replied to Coxon's post confirming Anthropic does "earnestly believe AI could kill all humans" and puts the odds at greater than 10 percent within a decade, per Morningstar. Hubinger added that Anthropic "is trying its best" but does not have a plan and is "not clearly on track" to prevent that outcome.

The obvious counterpoint

There's a fair question underneath all of this that deserves airing: when the CEOs of the two dominant AI labs, both preparing for enormous IPOs, simultaneously call for government-mandated safety rules and coordinated industry standards, that framework also raises the barrier to entry for smaller competitors who can't afford compliance armies. A regulatory structure built around "independent evaluators with employee-like access" sounds reasonable on its face, but it's also exactly the kind of rule that entrenches whoever is already biggest. No source in this reporting suggests that's the motive, and Amodei, Altman, and the departing researchers have all framed their concerns in terms of catastrophic risk, not market position. But the fact that Amodei's push for federal involvement lands in the same week as reports of a $2 trillion valuation target is a coincidence worth watching, not dismissing.

Greg Martin told MarketWatch that the attention around these resignations could push politicians to speed up AI regulation, saying, "We've been living in a pretty unregulated world as it relates to AI, and it could pressure them to start thinking about safety faster." What happens next depends on whether that pressure translates into actual rules, or whether Benton's and Coxon's warnings end up mostly as talking points in a fight over who gets to write the rules first.