The AI warnings are coming from inside the house

Shannon May

It’s as if a bunch of AI experts just had the same nightmare, because this week, several of them—including ones at Anthropic and OpenAI—seemingly jolted upright in a cold sweat to speak on the horrors they see coming from artificial intelligence. Anthropic: The company behind Claude lost its head of Safeguards Research, who announced his resignation in a letter that mentioned a world “in peril.” Speaking vaguely about Anthropic, he wrote, “Throughout my time here, I’ve repeatedly seen how hard it is to truly let our values govern our actions…we constantly face pressures to set aside what matters most.” At OpenAI, alarm bells came from three different employees this week: A researcher quit after two years due to “deep reservations” about ChatGPT’s new ad strategy, namely “a potential for manipulating users,” she wrote in an essay for the New York Times. A top safety executive was fired after opposing the upcoming release of AI erotica on ChatGPT, the Wall Street Journal reported. OpenAI said she was canned for sexually discriminating against a male coworker, which she called “absolutely false.” In a post on X that alluded to widespread job loss, an engineer wrote, “I finally feel the existential threat that AI is posing.” HyperWrite: The co-founder of an AI writing tool startup warned in a viral post on X that the latest AI models will render countless jobs obsolete, comparing the current moment to the weeks before the Covid-19 pandemic. How about some AI restrictions for the table? OpenAI and Anthropic are of two minds on how much to regulate their industry: Anthropic pledged $20 million to a political group that backs congressional candidates who favor AI safety, the company announced yesterday. (That’s substantial, but the company also announced yesterday that it closed a $30 billion fundraising round that valued it at $380 billion.) OpenAI has supported Leading the Future, a pro-AI super PAC that spends against the types of candidates that Anthropic’s donation would help. Meanwhile…half of xAI’s founders have exited as of this week, though the recent departures didn’t specifically cite AI concerns.