Enterprise & Security

Leaked Slack: OpenAI Staff Warned 'Leading the Future' Would Cost Them Researchers to Anthropic

2 min read AI-generated

The warning collected 24 agree reactions and a single eyes emoji. Brockman's retreat ten days later got 40 hearts.

Featured image for "Leaked Slack: OpenAI Staff Warned 'Leading the Future' Would Cost Them Researchers to Anthropic"

Semafor got hold of Slack messages from inside OpenAI and used them to reconstruct three days in June. Garrison Lovely published the piece on Friday.

What was in Slack on June 1

Jason Kwon, OpenAI’s chief strategy officer, showed employees a preview of a blog post. In it, the company denied any connection to Leading the Future, a super PAC set up to defend AI against regulation. President Greg Brockman’s support, Kwon wrote internally, “has been in a personal capacity, not on behalf of the company.”

Not everyone bought it. One employee went into Slack and named the price: the company’s aggressive political activity threatened to drive researchers to Anthropic — and so hamstring OpenAI in the race toward recursively self-improving AI. The post picked up 24 agree reactions and a single eyes emoji.

The next day Kwon gave ground to a different internal critic. The company was “taking reputational hits,” he wrote. They had drawn “some clear public lines,” but “I wish we’d done it even sooner.”

Ten days later, Brockman pulled out

Brockman backed out of the second half of his publicly pledged $50 million. He felt “terrible about the fact that LTF keeps reflecting on the company,” he wrote — a post the New York Times reported first. The Slack reaction: 40 hearts, five checkmarks, three thank-you emojis.

The two-camps picture doesn’t hold

From outside, the frontier labs split neatly: OpenAI the accelerationist cousin, Anthropic the safety-minded one. These Slack messages say otherwise. Plenty of OpenAI’s developers, engineers and researchers — exactly the people the company is fighting to keep in a talent war that also includes Meta, Google and a long list of startups — are not fine with unregulated AI.

What interests me is the argument that worked. Not ethics. Attrition. And it moved something that principle apparently wouldn’t have: ten days after that post, Brockman’s second tranche was gone. For Anthropic, that’s the uncomfortable punchline — it shows up here not as a moral authority but as leverage in a salary negotiation.

Sources:

OpenAIAnthropicPolitikRegulierung