Ex-Anthropic researcher's doom thread is now a copy-paste meme

Jacob Coxon quit Anthropic on Sept. 8. He had worked at the company for about four months, after a stint at OpenAI, and he left two months before his equity would have vested. His resignation post on X said AI labs are "racing straight to self-improving superintelligence and gambling with our lives." He also claimed that the people building AI privately believe it could kill us all by the end of the decade. The post passed 150 million views.

It didn't stay serious for long. Within a day, tech workers had turned it into a copy-paste meme. They kept Coxon's grave resignation tone and swapped the apocalypse for something absurd. Heatmap reporter Jae Holzman announced he was quitting over "self-painting fake tunnel technology," a Looney Tunes joke. Ramp's Ian Tracey warned about "self-closing books." Air CEO Shane Hegde raised the alarm on self-organizing workspaces. Others pasted in pop lyrics.

The meme didn't kill the debate, though. Anthropic alignment lead Evan Hubinger replied publicly and agreed with Coxon. He puts the odds of AI killing humanity within the next decade at more than 10%, and he said Anthropic doesn't yet have a plan to align superintelligence.

Why it matters: the internet's fastest defense against an existential warning is a punchline, but the insiders who agreed with Coxon aren't joking.