ChatGPT maker OpenAI failed to disclose a 2023 hack of its internal messaging systems, through which secrets of its technologies were plundered, reports the New York Times. A hacker gained access to internal discussions among employees, who shared details of OpenAI's latest artificial intelligence tech, in April 2023, but failed to access the company's AI systems, the Times reports. Company executives informed board members and employees last April but chose not to make the breach public since, as Reuters reports, "no information about customers or partners had been stolen."
Nor did OpenAI inform federal law enforcement. The company reportedly suspected a private individual, rather than a foreign government. Yet some employees were left fearing China or another foreign adversary could easily steal AI technology and use it against the US, per the Times. In a memo to the company's board, an OpenAI technical program manager reportedly claimed the company wasn't doing enough to prevent states from stealing its secrets. The manager was later fired, though OpenAI said that had nothing to do with the security concerns he raised.
"While we share his commitment to building safe [artificial general intelligence], we disagree with many of the claims he has since made about our work. This includes his characterizations of our security, notably this incident, which we addressed and shared with our board before he joined the company," a rep said. Even if the technologies were stolen, "there is not much evidence that today's AI technologies are a significant national security risk," per the Times. Some argue China doesn't need to steal US AI technologies with Chinese companies advancing by leaps and bounds on their own. Almost half of the world's top AI researchers hail from China, the Times reports. (More OpenAI stories.)