OpenAI Fires Three Safety Researchers Over Signal Chats

OpenAI has dismissed three artificial intelligence safety researchers following an internal security investigation into their communications with an external AI safety researcher. According to The Wall Street Journal, the company questioned the employees last month over suspected leaks involving confidential information about the internal designs of its AI models.

During the investigation, OpenAI's security team asked all three researchers to unlock their personal phones. Security officers examining the devices found that some of the researchers had enabled disappearing messages in their Signal conversations with the external researcher, according to The Wall Street Journal.

The company subsequently summoned the three employees to separate meetings scheduled simultaneously for 2 p.m. A week after the initial phone inspections, the researchers were informed of their dismissals.

Tomek Korbak, one of the researchers who lost his job, described the reason given to him by OpenAI as a breach of trust involving communications outside approved channels.

"I was told that there was a breach of trust because I interacted with a third party through unsanctioned channels," said Tomek Korbak, one of the fired researchers.

The investigation centered on suspected disclosures to an external AI safety researcher concerning the internal designs of OpenAI's artificial intelligence models. The Wall Street Journal identified those designs as among the company's most closely guarded secrets.

The phone inspections and subsequent dismissals highlight the company's handling of communications between employees and outside researchers when confidential AI development information may be involved. The available account does not establish whether the suspected disclosures were confirmed or specify what information, if any, was ultimately shared.