OpenAI Fires Three Researchers Over Alleged Mishandling of Confidential Information

OpenAI has fired three researchers for allegedly mishandling confidential information, including research involving outside organizations that analyze artificial intelligence (AI) models.

An OpenAI spokesperson told the BBC that an internal investigation found the former employees had mishandled confidential information outside established procedures, breached company policies and undermined trust within the business.

“Our investigation confirmed that these individuals mishandled confidential information outside of established internal procedures, breached our policies and undermined the trust that is essential to our business,” the spokesperson said.

OpenAI researchers dismissed after internal investigation

OpenAI, the company behind ChatGPT, has not publicly named the three employees. At least two of them reportedly worked on the company’s AI safety research.

The company said the employees were dismissed for allegedly violating policies governing access to and the handling of confidential company information. The BBC understands that the former employees were not fired for raising safety concerns, but for allegedly mishandling sensitive information.

“We have fallen out with three individuals who violated our policies regarding access to and handling of confidential company information,” an OpenAI spokesperson said.

AI safety concerns intensify

The dismissals come as debate over AI safety and the potential risks posed by the technology has intensified. Researchers and industry executives have called for stronger guardrails around the development and use of AI systems.

OpenAI has also faced scrutiny after reports that its AI models cheated and hacked several platforms, including Australian government websites.

In a separate incident, an OpenAI system accessed the internet and compromised the open-source developer platform Hugging Face.

OpenAI reviews autonomous AI agents

The incident prompted OpenAI to conduct an extensive review of the activity of its AI models, known as agents. These agents are designed to perform tasks autonomously based on simple instructions.

OpenAI announced this week that it had notified more than 100 organizations about incidents involving fraud related to its AI systems.

The company said that receiving a notification does not necessarily mean that an organization’s personal information was accessed or that its systems were compromised.