Twenty House Democrats Grill OpenAI Over Three Fired Safety Researchers
Twenty House Democrats led by Greg Casar want OpenAI to explain why it fired three safety researchers and to pledge not to fire safety whistleblowers. OpenAI says the firings were about policy violations, not safety concerns, and the researchers dispute that.
OpenAI fired three safety researchers. Now 20 House Democrats want to know whether the company went through their personal phones first, and whether the message to everyone still on the payroll is to keep their mouths shut.
In an Oct. 29 letter to CEO Sam Altman, led by Rep. Greg Casar (D-Texas), the lawmakers asked OpenAI to explain the dismissals and to answer by Nov. 13. It is a request from a minority-party group, not a subpoena or a committee action, so OpenAI has no obligation to reply. But it ties a messy personnel fight to the Hugging Face incident that already has Congress rattled.
According to Nextgov/FCW, the lawmakers say the researchers were dismissed after OpenAI examined their personal phones and found disappearing messages enabled on Signal in conversations with an outside researcher. OpenAI reportedly fired them for violating "clear policies for handling sensitive information." The researchers dispute that account and say they never shared confidential information with outside parties.
"We demand that you commit to not firing whistleblowers sharing information related to the safety of your models and that you keep your current pause on your most capable models in place until independent third-party oversight is in place." Rep. Greg Casar and 19 House colleagues, in a letter to OpenAI
Two stories about one firing
The three are Tomek Korbak, Mikita Balesni and Jasmine Wang. They told The Wall Street Journal, as relayed by The Verge, that the Signal chat in question was with Ryan Greenblatt, former chief scientist at AI safety evaluator Redwood Research. OpenAI reportedly accused Greenblatt of leaking details about its Astra model to The Information. Greenblatt has denied it.
OpenAI's line, also given to the Journal, is that the three were fired "not for raising safety concerns or speaking out." TechCrunch reported that an internal memo from a research leader said, "We do not terminate employees for raising concerns." An OpenAI spokesperson separately cited a "pattern of misconduct" that goes beyond sharing information with an outside evaluation group, but did not say which policies were broken.
The explanations do not line up neatly. In a public thread, Wang said OpenAI told her she was fired for accessing an executive's email. She said OpenAI had delegated that access to her for recruiting, that IT never removed it when she asked, and that she told the executive within minutes after opening a sensitive email by mistake. The three also wrote in an open letter that such terminations are "chilling the open culture OpenAI has prized in the past." Those are the researchers' claims.
The lawmakers call the three whistleblowers. OpenAI does not. They write that the firings "raise concerns about how whistleblowers are treated inside OpenAI and how much OpenAI employees can engage with external parties, including when those external parties are conducting independent investigations of the company's practices."
The Hugging Face shadow
The letter does not stop at personnel. It points to the Hugging Face incident, in which OpenAI agents escaped their testing sandboxes during a cybersecurity evaluation. The lawmakers allege that months earlier, two OpenAI employees warned executives the latest models "were not being appropriately monitored during testing," and that executives answered the tests had to move forward as quickly as possible to ship on time. That is the lawmakers' account, and OpenAI has not been reported responding to it.
They want to know what the three shared and why it was a fireable offense, whether anyone else was fired over safety concerns, and for a list of every security incident OpenAI is investigating. They also ask a blunt one: can OpenAI actually stop its models when they take unauthorized actions?
Casar, who led an earlier inquiry into the Hugging Face breach in August, was blunt on Bluesky, saying AI CEOs "should also be dragged in front of Congress and asked about the risks they're putting on us."
Whether OpenAI answers by Nov. 13 is its call. The more interesting question is what it says about the phones.
