Jacob Coxon pledges his OpenAI earnings to AI security

Jacob Coxon, a former researcher at OpenAI and then at Anthropic, committed on Wednesday to sell his OpenAI shares and dedicate the proceeds to AI safety causes. This promise comes after the hesitations he publicly expressed at Stanford.
At Stanford, Coxon was still hesitant to sell
On November 9, during a conference at Stanford, Jacob Coxon compared his stake in OpenAI to "contaminated blood." However, he was still reluctant to part with it: "I love money, man," he stated.
He also mentioned his parents, who, according to him, feared a psychotic crisis. In his view, giving up all his money would have heightened their concern.
Millions of dollars remain hypothetical
His stake could earn him millions of dollars during an IPO of OpenAI. Sam Altman, the company's CEO, has described this operation as imminent.
Le brief IA que les pros lisent chaque soir
Les 7 actus IA du jour, décryptées en 5 min. Gratuit.
Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.
Choisissez votre rythme
Gratuit · Pas de spam · Désabonnement en 1 clic
On X, Coxon is calling for a pause, or at least a marked deceleration, in the race towards superintelligence. He accompanies this request with a phrase: "Everything else is a distraction."
Coxon calls for a pause in certain developments
Originally from England and 27 years old, Coxon was employed by OpenAI from 2023 to 2026, and then by Anthropic, from which he left in September. His departure and warnings have sparked strong reactions, especially when he accused both companies of putting human lives at risk. In a message posted on X on September 8, he claimed that AI designers believed it possible that this technology could kill humanity before the decade ends, and that this was not, in his view, a marketing ploy. This message has garnered 175 million views.
Since then, he has been regularly speaking in the media to advocate for protections regarding AI. His request mainly focuses on a suspension of self-recursive improvement, which refers to the process where models train themselves.
Brief IA — L'actualité IA en français
L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.