OpenAI fires Jasmine Wang, two others over safety leaks

The terminations arrive as OpenAI and its rivals face growing pressure to be more transparent about safety incidents. OpenAI has been criticized for delays in disclosing incidents, including one where its agents hijacked a German wiki site — something the company reportedly knew about for months but failed to disclose until independent researchers revealed it. OpenAI has conceded it needs clearer standards for disclosure.

The firings also come amid a broader shift: last month, Anthropic said it would let outside evaluators verify its safety practices, and OpenAI opened its doors to outside researchers for six days after its agents hacked Hugging Face. Regulators are stepping up too — California Attorney General Rob Bonta issued an investigative subpoena to OpenAI, the FTC launched an investigation into AI safety practices at Anthropic and OpenAI, and last week Anthropic, OpenAI, Google, Meta, xAI, and Nvidia signed a voluntary White House pact to partner with independent external auditors.

Beyond the firings, David Robinson, who led the writing of safety reports for OpenAI's major product launches, resigned and published an essay arguing the company's culture is broken and that real pressure to prioritize safety must come from outside. Whether the fired researchers qualify as whistleblowers hinges on what information was shared and with whom — details that remain unclear — and on how OpenAI responds to congressional pressure. For now, the episode underscores that AI labs largely decide what the public learns about their safety incidents.