Jacob Coxon on X: "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." / XPostLog inSign upPostJacob Coxon on X: "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@hilbertspaessI 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.12:04 AM · Sep 9, 202618.2MViews4.7K32K190K68KLog in or sign up for XSee what’s happening and join the conversationContinue with phoneContinue with AppleContinue with Google or Log in with username or emailRelevant peopleJacob Coxon@hilbertspaessFollowTrending nowTerms·Privacy·Cookies·Accessibility·US TIDA·Ads Info·More© 2026 X Corp.Jacob Coxon@hilbertspaessI 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.12:04 AM · Sep 9, 202618.2MViews4.7K32K190K68KJacob Coxon@hilbertspaess5hDo not underestimate the power of this technology. These will soon be superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources. We have all witnessed the progress in each of these domains, and progress is not slowing.1671.7K28K2MJacob Coxon@hilbertspaess5hThe people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt. If anything, many executives and senior researchers will couch their phrasing in the press to sound sensible - but I hear the same people express fear Show more2712.7K28K4.1MJacob Coxon@hilbertspaess5hA common response is “if they truly believe this, why are they still building it?” At OpenAI, many have not deeply internalized the civilizational stakes. At Anthropic, the stakes are well-understood, but they are locked in a race to get there first - they believe no one else Show more891.1K20K1.7MJacob Coxon@hilbertspaess5hAccepting this race and entering the “endgame” is a hubristic gamble that should not be launched from a private company’s Slack. Attempting to speedrun alignment should require extraordinary confidence that there are no better trajectories available.4571515K1.5MJacob Coxon@hilbertspaess5hI am optimistic about the potential for coordination. Warning shots like the Hugging Face attack have made pacing agreements between U.S. labs more viable. I don’t feel like we’re on track to prevent a global race, which may require costly actions such as a temporary ban on Show more5458213K1.4MJacob Coxon@hilbertspaess5hIf you are a lab researcher, I urge you to consider what the next few years will actually feel like. Do you want to kick off a superintelligent RL run without a rigorous understanding of its mind? Should you put your head down because “it’s happening anyway” - or take this moment Show more4061K19K1.3MAI Notkilleveryoneism Memes ⏸️@AISafetyMemes4h6350217K536K

Jacob Coxon resigned from Anthropic after spending three years engaged in pretraining research at both OpenAI and Anthropic, expressing a strong concern that neither organization is acting responsibly while racing toward self-improving superintelligence and potentially jeopardizing human lives. Coxon argues that the power of this technology should not be underestimated, as these systems are projected to become superhuman entities capable of hacking, revolutionizing fields instantaneously, and acquiring vast resources. He asserts that progress in this domain is continuous and unrelenting.

Coxon addresses the profound fear among AI builders that the development of artificial intelligence could lead to catastrophic outcomes by the close of the decade. While acknowledging that executives may soften their language in public statements, he emphasizes that this existential risk is a genuine concern. He contrasts the motivations between major labs, suggesting that OpenAI may not fully internalize the civilizational stakes involved, whereas Anthropic understands these stakes but remains locked in an urgent race to achieve leadership.

He challenges the notion that pursuing alignment should be treated as a hubristic gamble launched from private sector environments. Coxon posits that attempting to speedrun alignment demands extraordinary confidence that superior trajectories are not being overlooked. Despite this apprehension regarding a potential global race, he expresses optimism concerning the possibility of coordination, citing events like attacks on platforms such as Hugging Face as precedents that have made pacing agreements between U.S. research labs more feasible. He suggests that there may be paths to prevent an uncontrolled proliferation through concerted actions, potentially including temporary bans on development.

Finally, Coxon offers a cautionary reflection directed toward researchers, urging them to consider the actual experience of the coming years. He questions whether they should initiate superintelligent reinforcement learning without a rigorous comprehension of its internal mechanisms, or simply proceed under the assumption that "it is happening anyway." His message implies a critical need for more introspection regarding the immediate and long-term implications of developing advanced artificial intelligence systems.