Former Anthropic researcher Jacob Coxon will testify at a New York City hearing on artificial intelligence. Bloomberg News reported the plan on Sunday, citing people familiar with the matter.
The hearing is tied to work by New York City lawmakers. They are reviewing a package of bills aimed at setting up safeguards around AI technology.
Coxon worked as a researcher at Anthropic, the company that builds the Claude AI models. He left the company last month.
Why Coxon Is Testifying
Coxon will appear at the request of New York City Council Speaker Julie Menin. That is according to the Bloomberg report.
Menin has urged AI whistleblowers to come forward and testify. She made the call as the City Council considers the proposed AI bills.
The report did not give a date for the hearing. It also did not list which specific bills will be discussed.
Details about what Coxon plans to tell lawmakers were not included in the report. The people familiar with the matter were not named.
The package of AI bills is still being weighed by city lawmakers. No vote on the bills was mentioned in the report.
Warnings After Leaving Anthropic
Coxon quit Anthropic in September. When he left, he issued public warnings about the dangers of AI.
He said the "people building AI earnestly believe that it could kill us all by the end of the decade." He made the statement after leaving the company.
Coxon also accused his former employer and OpenAI of "gambling with our lives." OpenAI is the company behind ChatGPT.
Anthropic and OpenAI are two of the largest AI developers in the United States. Both companies build advanced AI models used by consumers and businesses.
The report did not include a response from either company. It is not clear if Anthropic or OpenAI will take part in the hearing.
Coxon's appearance would place a former AI industry insider in front of city lawmakers. He would be testifying as the council reviews rules for the technology.
Reuters published its story on the Bloomberg report on Sunday, October 4. The news agency said it could not immediately verify the report.
