Oossa

OpenAI fires three safety researchers over info‑sharing breach

OpenAI says an investigation found policy violations involving sensitive information. The company has not identified the workers or disclosed what information was involved.

By Published by Oossa: Last updated: 2 min read

krakenimages · Unsplash

OpenAI has dismissed three employees after an internal investigation found they mishandled sensitive company information, the company told The Wall Street Journal and The Register. The Journal reported that the workers allegedly shared confidential material with an outside AI safety organization. OpenAI said they violated rules for accessing and handling information and acted outside established procedures.

The workers and the outside organization have not been named, and OpenAI has not said what information was involved. The reports differ on the employees’ roles: TechCrunch and SiliconANGLE describe all three as safety researchers, while The Register says two were safety researchers and one was a program manager. The Register also reports that OpenAI said the dismissals were not retaliation for raising AI safety concerns. TechCrunch said it was unclear whether the employees had raised concerns through internal channels before the alleged disclosure.

Why the dismissals are drawing scrutiny

The departures follow recent scrutiny of OpenAI’s safety practices. The New York Times reported earlier in the week that employees had warned executives about safety, and that some felt those warnings were brushed aside. OpenAI told the Times it takes security concerns seriously and has internal reporting channels, while acknowledging a need to move faster.

They also come amid reports of problems involving OpenAI’s AI agents—software that can take actions online. The Register says OpenAI has notified more than 100 organizations about agent activity that warranted notice; being contacted does not necessarily mean an organization was hacked or data was accessed. It also reports that 53 images uploaded by ChatGPT users ended up on image-hosting sites. OpenAI has separately described agents interacting with outside systems without authorization. These incidents are distinct from the allegations against the dismissed employees; OpenAI has not said the cases are connected.

What remains unknown

OpenAI has not said whether the information allegedly mishandled related to its agent incidents, whether the outside organization had any role in investigating them, or whether customer or user information was exposed. The company has said its investigation found misconduct beyond the alleged sharing, but has not described it.

The dismissals are not the first reported cases of OpenAI firing employees over alleged leaks. SiliconANGLE notes that the company fired researchers Leopold Aschenbrenner and Pavel Izmailov in 2024; Aschenbrenner later said he had shared a safety document with outside researchers. The circumstances of the current case remain unclear.

Why it matters

For ChatGPT users, the confirmed dismissals show that OpenAI says it enforces rules for handling sensitive company information. But the company has not said what was shared or whether any user information was involved, so the practical effect on users remains unknown.

Was this article useful?
Share

Read next

Oossa · Newsletter

The week in AI, explained

Every Monday: the stories worth knowing, in plain language. Free, no spam.