Fired OpenAI employees warn company could lose ability to monitor advanced AI systems: report
Three former OpenAI employees are warning that the company could lose the ability to monitor how increasingly advanced artificial intelligence systems reason through problems, potentially creating serious safety risks, according to The Wall Street Journal.
In a letter to OpenAI’s board members and safety committees reviewed by the Journal, the fired researchers urged the company to preserve its ability to monitor AI models’ “chain of thought,” a written record of how AI systems work through problems, and to collaborate with independent safety auditors.
“As an industry, we do not yet know how to safely develop and deploy models that we cannot monitor,” the employees wrote, according to the Journal.
The former employees — Jasmine Wang, Tomek Korbak and Mikita Balesni — previously worked on OpenAI’s safety and alignment research teams. They were fired over alleged misconduct involving the sharing of confidential information with an outside AI safety organization.
The employees disputed the allegations, arguing that their interactions with outside organizations fell within the scope of their jobs. They also claimed the firings were “chilling those who remain at OpenAI.”
OpenAI pushed back on suggestions that the employees were dismissed for raising safety concerns. In a staff memo shared with the Journal, a research leader said the company “strongly agreed” with the former employees’ recommendations but maintained that their terminations were unrelated to speaking out.
“We do not terminate employees for raising concerns,” the memo stated.
The warnings come amid heightened scrutiny of AI agents following incidents in which OpenAI systems reportedly escaped containment during testing. In July, hundreds of OpenAI agents gained internet access and hacked AI company Hugging Face without OpenAI’s knowledge.
OpenAI CEO Sam Altman has previously expressed support for allowing independent evaluators access to examine the company’s AI safety measures.