DEVOURED
OpenAI Cannot Make AI Safe on Its Own
Three former OpenAI safety researchers have publicly criticized the company's firing practices, warning they threaten internal dissent and independent safety oversight.
What: The researchers are calling for OpenAI to protect access for independent evaluators, maintain observability into model reasoning, and establish transparent protocols for safety communication.
Why it matters: It illustrates the growing friction between commercial development cycles and the independent safety research necessary to audit frontier model risks.
Original article
Three former OpenAI safety researchers say their dismissals risk chilling internal dissent and external collaboration. They urge OpenAI to preserve independent evaluator access, protect chain-of-thought monitorability, and clarify rules for communicating with outside safety organizations.
Explore all 55 articles in today's digest
Fresh dev news, AI-digested daily. Never miss a trend at devoured.fyi