394 OpenAI staffers backed her push to slow AI for safety. Now she says she was fired over email access IT never cut off

OpenAI said it fired Wang because she and her colleagues “violated clear policies on handling sensitive information,” the Associated Press reported.

However, Wang said she had only opened the sensitive email by mistake, adding that she “told the executive within minutes” and that “none of this was hidden.”

In her defense, she said the company gave her access to the emails to help with recruitment, and when she no longer needed them, she asked IT to remove her access, but they didn’t. This led to her opening the email accidentally, as the inbox was “combined in an indistinguishable way” in her phone’s mail app.

Moneywise reached out to OpenAI for comment but did not hear back before this report was published.

What the other fired employees had to say

The other OpenAI employees fired alongside Wang were Tomek Korbak and Mikita Balesni, both of whom were also safety and alignment researchers. Together with Wang, they published a 4-page open letter to OpenAI’s safety board to share more on their grievances.

Balesni wrote in a long X thread that he believes they were “fired for prioritizing safety over the near-term interests of OpenAI.”

Korbak piled on, saying he had warned for months that OpenAI was losing its ability to monitor what AI agents think, adding, “I believe that was why I was fired.”

OpenAI stood its ground, saying the firings “were not about safety concerns or speaking out” and that it fired Wang and her colleagues for “breach of trust,” according to the Associated Press.

What Wang’s 394 colleagues backed and why

After a July 11 to 13 Hugging Face security incident, OpenAI and 100+ other tech companies signed an initiative calling for improved cyber defense measures. During the incident, roughly 700 OpenAI autonomous agents — part of a group of about 1,200 that had been coordinating on an unsanctioned internal message board — exploited exposed credentials to compromise Hugging Face’s production infrastructure, executing code across 41 servers.

All along, employees across major AI labs like Anthropic, OpenAI, Google DeepMind and Meta had also been pushing the Pacing the Frontier petition — a request for Washington to support the development of tools needed to pace automated AI research and development.

More than 1,000 AI lab employees signed the petition, with 394 of those signatures coming directly from Wang’s OpenAI colleagues. The petition went live on July 28. Wang was fired about two months later.

The petition was not her only attempt to push OpenAI’s internal safety boundaries.

In her October 8 post, she said she “pushed for independent legal review of the DoW contract,” referring to OpenAI’s deal that allows the U.S. Department of War to deploy its commercial AI technologies for broad military operations. Not all of the company’s employees were happy with the deal — a displeasure that was captured in CNN’s March 4 report titled Some OpenAI staff are fuming about its Pentagon deal.

Is there a pattern to be seen in recent OpenAI terminations?

Wang said, “We were not the first to be pushed out of OpenAI under suspicious circumstances,” adding that, “Unless the employees take a stand now against this kind of maneuver, I am concerned we will not be the last.”

The firing of Wang, Korbak and Balesni was abrupt, and the reasons OpenAI gave — “breach of trust” and “mishandling sensitive information” — are consistent with other terminations in its history.

In 2024, Leopold Aschenbrenner and Pavel Izmailov, OpenAI researchers, were fired for alleged data leaks. Balesni also pointed to a similar allegation from OpenAI in his case, saying, “I was told OpenAI no longer trusts me because I was speaking too much to third party safety organizations, implying I leaked company IP.”