Days after AI safety researcher Jacob Coxon quit from Anthropic warning that AI companies are gambling with people’s lives by racing towards superintelligent AI, he held an ask me anything (AMA) session on social media wherein he insisted that companies like Anthropic should walk the talk on AI safety.
Coxon, a Cambridge University AI safety researcher, who previously worked at OpenAI and later joined Anthropic in 2026, posted a viral tweet which started a debate on whether tech companies are aware of the inherent harms if AI reaches recursive self-improvement (RSI) i.e. a hypothized state where an AI model designs, develops and trains its own superior version.
Doubling down on the issue, Coxon answered an X user during the AMA session who asked where he disagrees sharply with his previous organization and what prevented him to address this issue internally, Coxon replied stating that his core disagreement with Anthropic was the “inevitability of RSI race”.
0. The core disagreement was about the inevitability of a race
1. I think leadership is way too paranoid about China and the US government. They don’t believe it will be possible to negotiate.
2. They largely initiated the recent race to RSI, because of a belief in its…— Jacob Coxon (@hilbertspaess) September 15, 2026
“They (Anthropic) largely initiated the recent race to RSI, because of a belief in its inevitability. Note that OpenAI had to shed a bunch of dead weight like Sora because Anthropic was going for the jugular. Even if they are **not** being pessimistic, I disagree with their consequentialist philosophy. If the race is inevitable you should not contribute,” said Coxon.
Coxon said he believed Anthropic’s leadership was “way too paranoid” about China and the US government and did not believe it would be possible to negotiate with them.
Also Read: Explained: Why is Anthropic Calling for a Global Pause on Frontier AI?
In another response to a question from a user on X about whether a super intelligent system could recognize its human creators and understand the values and principles given to it, Coxon said current AI models already demonstrate some understanding of their own origins and training.
He further said that a future super intelligent AI could understand that humans created it and could potentially understand what humans want from it, but that this would not necessarily ensure that the system would act according to human intentions.
Hi!
1. Yes, and indeed I think current models already understand this to a degree. They’ve certainly read about their training procedures on the internet. In fact, there’s some worry that we could “hyperstition” certain futures if superintelligence internalizes narratives we’ve…— Jacob Coxon (@hilbertspaess) September 15, 2026
The former Anthropic employee expressed that he also feels “disempowered” about the future of AI, following a question by Michaël Trazzi, founder and CEO of The Inside View, a media company producing documentaries about AI Safety and who is also currently leading movement called “StopTheAIRace”
Michaël asked Coxon in a post on X what he would say to the millions of people who had read his post and felt disempowered or hopeless about the future of AI. Coxon responded that a sense of hopelessness about the future was partly behind his decision to reign.
“I also feel disempowered. Part of resigning was a feeling of hopelessness about the future. I would say- keep your eyes open as things get crazier and advocate for increased transparency into AI companies,” Coxon tweeted.
I also feel disempowered. Part of resigning was a feeling of hopelessness about the future. I would say- keep your eyes open as things get crazier and advocate for increased transparency into AI companies..
— Jacob Coxon (@hilbertspaess) September 15, 2026
Meanwhile some of the users on X have also praised Coxon for taking such an extreme step and expressed their solidarity for him.
“I have no questions. Just wanted to make sure you’re doing okay man. Huge whirlwinds of attention, harassment, conspiracies etc like what you experienced are not easy to go through. Mad respect for you and how you handled all of this,” tweeted one of the users.
Coxon, a researcher from Cambridge University, who has worked at both OpenAI and Anthropic, took to X and wrote, “ I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives.
I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.
— Jacob Coxon (@hilbertspaess) September 9, 2026
Following Coxon’s tweet, concerns have grown over the potential risks posed by the rapid development of artificial intelligence. It has also intensified the political heat in the U.S. as several governors, senators, lawmakers, senior congressmen and congresswomen expressed their shock and concern.
Coxon received the most notable praise from senator Bernie Sanders who said, “The very people building this technology admit that it could threaten the future of humanity. That is why I will soon be introducing legislation to ban superintelligence and pause AI development.”
