I’ve had this same conversation about four times in the last two weeks: with clients, with people I just talk to about this stuff, and with a couple of folks who caught me after a call to ask what I actually thought. Someone brings up “the Anthropic guy who says AI has a 10% chance of ending humanity” and wants to know what’s real. So let’s walk through what actually happened, because there’s more underneath this than one viral tweet.
On September 8th, a researcher who’d been at Anthropic for only a few weeks resigned and posted a thread saying both Anthropic and OpenAI were racing toward self-improving superintelligence and gambling with our lives. That thread hit 100 million-plus views in 24 hours. The Wall Street Journal had a piece published about 18 minutes before his own thread went out. Whether that’s coincidence or coordination, I can’t tell you, but the speed and scale this thing spread at are real, and they matter before you take a side on any of it.
This didn’t happen in a vacuum, either. Back in 2021, the investors behind Anthropic’s Series A are largely the same people funding a lot of the outside “AI safety” research groups now publishing evaluations of these companies. In 2025, the FTC under Lina Khan started looking into these AI partnerships and cross-investments, and that effort stalled once Khan left the agency. There have also been a few attempts in Congress to set federal AI rules; none have gone anywhere yet.
An hour and a half after the resignation thread went up, Anthropic’s own alignment lead, the team responsible for making sure AI doesn’t do something dangerous, publicly agreed there’s a greater-than-10% chance of AI causing human extinction this decade. Four days later, Anthropic CEO Dario Amodei published a proposal called “Pacing the Frontier.” His ask wasn’t just “let’s slow down.” It was specifically for an antitrust waiver or a DOJ business review letter, so competing AI companies could coordinate on safety standards without risking a collusion charge. Sam Altman backed the same idea publicly. Elon Musk posted, simply, “Dario is right.” Three of the biggest names in AI, all agreeing the risk is real enough to slow down, but all saying they need the government’s permission to do it together.
That’s where it gets interesting, because the pushback came from two directions that don’t usually agree. The Trump administration, through AI adviser David Sacks and Trump himself, framed this as a regulatory capture play: convenient timing for companies who’d benefit from rules that make it harder for anyone smaller to catch up, dressed up as a safety concern. Trump’s own take was blunter: he called the whole thing a hoax, said he’s the only regulator AI needs, and that the real priority is staying ahead of China. Lina Khan came at it from the opposite direction. Her argument: we don’t need new AI-specific rules, because existing consumer-protection and antitrust law already covers companies that release dangerous products. She’s leaning on decades-old Supreme Court precedent that says competing companies don’t get a pass for cutting corners just because a rival might do the same thing.
Key Takeaways
- Tech leaders Amodei, Altman, and Musk think the AI risk is real enough to slow down together, but say they need government permission to do it without antitrust exposure.
- The Trump administration and Lina Khan both oppose new AI-specific rules right now, for completely different reasons: one says existing law already covers it, the other says the risk itself is overblown.
- There’s real, disclosed overlap between the investors funding AI companies and the investors funding the outside groups evaluating them, worth knowing before treating that research as fully independent.
- You don’t need to pick a side to be useful in this conversation. You need both arguments stated fairly.
Want the CPE credit? Take the full lesson on EverydayCPE and earn 0.2 CPE credits: https://everydaycpe.com/courses/technical-information-technology/the-ai-doomer-debate/
Leave a Reply