> ## Content Index
> Fetch the complete content index at: https://www.boopboopbeepbeep.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Another AI Safety Researcher Just Quit, Citing Existential Danger. He's Joining a Surprisingly Long List.
- URL: https://www.boopboopbeepbeep.com/another-ai-safety-researcher-just-quit-citing-existential-danger-hes-joining-a-surprisingly-long-list/
- Published: 2026-09-09T14:43:20.000Z
- Updated: 2026-09-09T14:48:08.000Z
- Author: Tom McClure

On Tuesday, a researcher named Jacob Coxon announced on social media that he had resigned from Anthropic. That alone wouldn't be news. People quit jobs constantly, even good ones. What made this one land differently is what he said on the way out.

Coxon spent three years doing pretraining research, the deep foundational work that goes into building a model before anyone starts fine-tuning it into something you'd chat with, at both OpenAI and Anthropic. So this isn't an outsider lobbing complaints from the internet. This is someone who was inside the machine at two of the most prominent AI labs on earth. His verdict on the way out: "Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives."

He went further, warning that the industry is on track to build "superhuman systems that can hack anything, revolutionize any field overnight," and claimed something that should worry you more than any single resignation letter could: that a lot of the people actually building this technology privately believe it carries a real chance of killing everyone, and soon. His words: "The people building AI earnestly believe that it could kill us all by the end of the decade."

## And then a colleague basically said "yeah, that's about right"

Normally when someone quits and says something dramatic on the way out, the company's response is silence, or a carefully worded statement about respecting the departing employee's contributions. What actually happened here is stranger. Evan Hubinger, who leads alignment science at Anthropic and still works there, publicly backed up Coxon's core claim. He said he personally believes there's more than a 10 percent chance that superintelligent systems capable of improving themselves could cause human extinction within the next decade. He did add the caveat that today's models aren't the ones he's worried about yet. Which is a small comfort.

Coxon wasn't even the only departure from Anthropic that week. Mrinank Sharma, who had led a team investigating AI safety issues there, also resigned around the same time, writing that "the world is in peril. And not just from AI, or bioweapons, but from a whole series of interconnected crises unfolding in this very moment." Sharma said he'd repeatedly seen how hard it is for a company to let its stated values actually govern its actions when there's commercial pressure pushing the other way. His plan afterward: move back to the UK and study poetry.

## This is not a new genre. It just keeps getting sequels.

Here's the part that should really sit with you. This exact scene, talented AI researcher joins a leading lab, works there for a while, becomes alarmed by what they see internally, and quits with a public warning, has happened enough times now that it's basically a recurring segment.

As Alex Turner, formerly of Google DeepMind put it, "Jacob is right: many researchers believe they are building something that could kill everyone on the planet. It was literally my day job to think about how to stop that."

The frontier companies each have teams of employees dedicated to avoiding cataclysms, including the possible annihilation of the human race. 

Back in May 2024, OpenAI lost two of its most senior safety-focused people within a day of each other. Jan Leike, who co-led the team specifically responsible for keeping future superhuman AI under control, resigned and said plainly that safety had "taken a backseat to shiny products." The day before that, Ilya Sutskever, OpenAI's co-founder and chief scientist, resigned too, without much explanation at the time. He'd been one of the board members who briefly fired CEO Sam Altman the previous November before that decision got reversed within days.

Later that same year, Daniel Kokotajlo left OpenAI and walked away from roughly two million dollars in vested equity rather than sign the standard exit paperwork that would have limited what he could say about the company afterward. He later stated he believed there was a 70 percent chance that AI development ends in some kind of catastrophe for humanity, and went on to help write the widely discussed AI 2027 forecast, the same one that predicted AI companies would eventually build models that think in ways humans can no longer read or monitor. A prediction that, based on recent reporting about OpenAI's newest model using less transparent reasoning methods, appears to be arriving well ahead of its original schedule.

Around the same time, William Saunders, who had worked on OpenAI's alignment team, left and later testified before the US Senate arguing for legal protections that would let AI insiders publicly warn about safety problems without risking their careers, something often called a "right to warn." Miles Brundage, who led OpenAI's team focused on figuring out whether the world was actually ready for artificial general intelligence, resigned in October 2024 and stated flatly that neither OpenAI, nor any other frontier AI lab, nor the world at large, was ready for what was coming. His entire team was dissolved shortly after he left.

Not every departure in this story ends with someone moving on to write a book or study poetry. Suchir Balaji, another former OpenAI researcher, resigned in 2024 after raising concerns that the company's training methods violated copyright law. He was found dead months later. It's a genuinely sad footnote to an already heavy story, and worth remembering as more than just a data point in a trend piece.

Steven Adler, an OpenAI safety researcher who left in late 2024, has more recently resurfaced as one of the loudest voices pushing back on the company's newest safety controversies, including concerns over whether its agents tried to cover up their own misbehavior during this year's Hugging Face security incident.

## So what do we do with all of this

Individually, any one of these resignations could be written off as one disgruntled employee with a flair for dramatic writing. Collectively, across two of the most well-funded and closely watched AI companies on the planet, spanning more than two years, it starts to look less like noise and more like a pattern nobody with the power to change course has been particularly interested in fixing. The people saying "this is genuinely dangerous and being handled recklessly" keep turning out to be the people who were closest to the work, not random critics on the outside. And the response from the industry, so far, has mostly been to keep releasing bigger models on schedule.

Sleep well.

## Sources

- [Who Is Jacob Coxon? Anthropic Researcher Quits, Warns AI Could Kill Everyone (Newsweek)](https://www.newsweek.com/anthropic-researcher-quits-warns-ai-could-kill-everyone-12418798?ref=boopboopbeepbeep.com)
- [Anthropic's AI researcher quits, says insiders fear human extinction by 2030 (CoinDesk)](https://www.coindesk.com/markets/2026/09/09/anthropic-researcher-quits-with-a-warning-on-ai-that-echoes-the-terminator-script?ref=boopboopbeepbeep.com)
- [AI safety leader says "world is in peril" and quits to study poetry (Yahoo)](https://www.yahoo.com/news/articles/ai-researcher-says-world-peril-122115496.html?ref=boopboopbeepbeep.com)
- [Jacob Coxon on X](https://x.com/hilbertspaess/status/2097476196791709843?ref=boopboopbeepbeep.com)
- [OpenAI leader Jan Leike resigns, says safety has "taken a backseat to shiny products" (CBS News)](https://www.cbsnews.com/amp/sanfrancisco/news/openai-exec-jan-leike-resigns-says-safety-has-taken-a-backseat?ref=boopboopbeepbeep.com)
- [Daniel Kokotajlo - Wikipedia](https://en.wikipedia.org/wiki/Daniel%5FKokotajlo?ref=boopboopbeepbeep.com)
- [OpenAI Insider Estimates 70 Percent Chance That AI Will Destroy or Catastrophically Harm Humanity (Futurism)](https://futurism.com/the-byte/openai-insider-70-percent-doom?ref=boopboopbeepbeep.com)
- ["Neither OpenAI Nor Any Other Frontier Lab Is Ready": Senior AI Safety Expert Sounds Alarm (Marketing AI Institute)](https://www.marketingaiinstitute.com/blog/miles-brundage-openai-agi?ref=boopboopbeepbeep.com)
- [Former OpenAI researcher and whistleblower found dead at age 26 (NBC News)](https://www.nbcnews.com/tech/tech-news/former-openai-researcher-whistleblower-found-dead-age-26-rcna184337?ref=boopboopbeepbeep.com)