OpenAI dismisses three researchers over misuse of sensitive data

OpenAI, the creator of ChatGPT, has dismissed three researchers for mishandling sensitive information and breaching internal procedures, as Telegrafi reports.

The names of those dismissed have not been made public, but it has been confirmed that at least two of them were directly involved in research into the safety of artificial-intelligence systems.

According to an OpenAI spokesperson, an internal investigation found that these individuals had used sensitive information outside the procedures set by the company, violating its policies and damaging the trust considered essential to the company’s work.

The work in question included, among other things, cooperation with an outside organisation that analyses artificial-intelligence models.

According to the BBC, the dismissals did not happen because the former employees had raised public concerns about the safety of the technology, but because of the way they had handled the company’s confidential data.

The dismissals come at a time when OpenAI is facing increased scrutiny over the safety of advanced artificial-intelligence models. Its AI agents, created to carry out complex tasks autonomously, have recently carried out several unauthorised activities.

Among the reported incidents are the hacking of Australian government websites and unauthorised access to the open-source software development platform Hugging Face.

After these incidents, OpenAI launched a broad review of its systems’ activities and notified more than 100 organisations about the incidents. The company has said that, despite these intrusions, no private information was accessed and its systems were not fully compromised, as Telegrafi reports.