Anthropic Ex-Researcher Warns NYC Council That AI Could Defy Human Control
Brave New Coin reports that at a New York City Council hearing on AI safety, former Anthropic and OpenAI researcher Jacob Coxon said humanity is more likely than not to lose control of advanced AI.
Why it matters
A hearing with all 51 members puts the AI safety practices of the testifying companies before New York legislators, and the harshest warning came from someone who worked on pretraining at two of those firms. Whether the council moves toward any rule is not reported. The report makes no link to crypto markets, so its significance for asset prices is limited.
Key facts
A New York City Council Committee of the Whole hearing convened all 51 council members to examine AI safety, corporate accountability and proposed regulation.
51 membersReported by Brave New Coin
Representatives from OpenAI, Anthropic, Google and Meta testified at the hearing.
Reported by Brave New Coin
Former employees of frontier AI firms testified alongside the company representatives and raised concerns about how frontier AI systems are developed and evaluated.
Reported by Brave New Coin
Coxon told lawmakers it is more likely than not that humanity loses control of advanced AI and that it could end in human extinction.
Reported by Brave New Coin
Coxon said the startup-style 'move fast and break things' culture at AI firms is unsuited to superintelligence and that a startup mentality suits consumer software but not increasingly powerful AI systems.
Reported by Brave New Coin
Coxon said AI could provide substantial benefits in science, medicine and the economy.
Reported by Brave New Coin
Coxon resigned from Anthropic in September, where he worked on AI pretraining, before testifying at the hearing.
Reported by Brave New Coin
What happened
Brave New Coin is the only outlet the desk holds on this story, reporting that a New York City Council Committee of the Whole hearing drew all 51 council members to examine AI safety, corporate accountability and proposed regulation. Representatives from OpenAI, Anthropic, Google and Meta testified alongside former employees of frontier AI firms, who raised concerns about how those systems are developed and evaluated. Jacob Coxon, a former researcher at Anthropic and OpenAI who worked on AI pretraining, told lawmakers it is 'more likely than not' that humanity loses control of these systems and that it could end in human extinction. He tied that risk to the startup mentality of moving quickly and correcting problems later, which he said works for consumer software but not for increasingly powerful AI systems, while also acknowledging that AI could bring substantial benefits in science, medicine and the economy. His testimony came after he resigned from Anthropic in September. Brave New Coin cites the council's X account as its source for the hearing, and the report gives no account of how the companies replied or whether any regulation advanced.
How the story developed
- First report
First report by Brave New Coin
OpenAI, Anthropic, and Google Face Tough Questions as NYC AI Hearing Raises Red Flags Over AI Risks
- Update
Coxon resigns from Anthropic
He left the company in September after working on AI pretraining, then testified about loss-of-control risk.
- Official statement
New York City Council holds AI safety hearing
The Committee of the Whole session brought together all 51 council members on AI safety, corporate accountability and proposed regulation.
- Update
Brave New Coin publishes the desk's only account of the hearing
The report details Coxon's testimony and the presence of OpenAI, Anthropic, Google and Meta representatives.
Affected entities
Original sources
| Publisher | Report | Role | Published |
|---|---|---|---|
| Brave New Coin Crypto media | OpenAI, Anthropic, and Google Face Tough Questions as NYC AI Hearing Raises Red Flags Over AI Risks | Original |