OpenAI Security Breach Revealed by The New York Times
OpenAI experienced a security breach in 2023 but did not disclose the incident outside the company, as reported by The New York Times on July 4.
OpenAI executives disclosed the incident internally during an April 2023 meeting but opted not to reveal it publicly as the attacker did not access sensitive information about customers or partners.
Executives did not view the incident as a national security threat due to the attacker being a private individual without ties to a foreign government. They did not report the breach to law enforcement.
The attacker infiltrated OpenAI’s internal messaging systems and obtained details about the firm’s AI technology designs from employee conversations on an online forum. They did not breach the systems housing AI technology or access any code.
Sources cited by The New York Times provided these details.
Concerns Raised by Ex-Employee
Former OpenAI researcher Leopold Aschenbrenner expressed concerns in a memo to OpenAI directors post-incident, urging measures to prevent foreign entities from stealing company secrets, as referenced by The New York Times.
Aschenbrenner mentioned the incident on a recent podcast episode.
OpenAI spokesperson Liz Bourgeois acknowledged Aschenbrenner’s concerns and affirmed support for safe AGI development while disputing certain claims. She stated, “We disagree with many of [Aschenbrenner’s claims] … This includes his characterizations of our security, notably this incident, which we addressed and shared with our board before he joined the company.”
Aschenbrenner claimed he was terminated from OpenAI for leaking information and political reasons. Bourgeois refuted this, stating his concerns did not lead to his dismissal.
OpenAI’s head of security, Matt Knight, emphasized the company’s security efforts and commitment. He mentioned that they had been investing in security long before ChatGPT and acknowledged the risks associated with AI development that need to be managed.
The New York Times disclosed a potential conflict of interest by noting their legal battle with OpenAI and Microsoft over alleged copyright infringement. OpenAI believes the lawsuit lacks merit.
