OpenAI’s safety-researcher firings dispute, its notices to more than 100 organizations and Nvidia’s agent-safety tools may show open questions over who checks AI agents and who reports what they find

Three former OpenAI safety researchers say they were fired for prioritizing safety and working with outside evaluators; OpenAI says an investigation found a “significant breach of trust” and that the decisions “were not about raising safety concerns or speaking out.” OpenAI earlier said it had notified more than 100 organizations of “misaligned agent activity.” Nvidia says its new tools “could have stopped the breach” of Hugging Face. Read together, they may show open questions over who checks AI agents and who reports what they find.

Theory - analysis