Joe Benton leaves Anthropic and calls for independent safety oversight

Joe Benton said he had left Anthropic’s safety team two weeks earlier and would join METR. He argued that AI companies were underinvesting in safety and that the public needed more visibility into their progress and risks. September 11 is the statement date; the post gives only a relative date for his departure. Joe Benton 

He called for disclosure of progress toward recursive self-improvement, incident and near-miss reporting, minimum safety standards and independent checks. His biography identifies his former role managing Anthropic’s Scalable Oversight team. Joe Benton Joe Benton