CSO:
A previously unreported security breach at OpenAI, the developer of ChatGPT, has raised alarms over the potential of foreign adversaries, such as China, accessing sensitive AI technologies.
While the hacker did not access the core code, the incident that took place in 2023 has sparked fears over foreign adversaries — particularly China — stealing sensitive AI secrets, The New York Times reported.
The breach occurred early last year when a hacker infiltrated OpenAI’s internal messaging system, gaining access to employee discussions regarding the company’s latest AI advancements, the newspaper said quoting two anonymous sources.
OpenAI, headquartered in San Francisco, confirmed the breach to employees as well as the board of directors in April 2023 in an all-hands call but opted not to make it public. The company reasoned that no customer or partner data was compromised, and they believed the hacker was an individual, not a state-sponsored actor, the report added.
OpenAI executives opted not to publicly disclose the incident, believing the stolen information posed no immediate threat and the hacker was likely a lone actor. The company also did not inform the law enforcement agencies including the FBI, the two sources told the newspaper.
This incident, however, raised concerns among OpenAI employees that foreign adversaries such as China could steal the company’s AI technologies and may threaten US national security, the newspaper added.
After the said breach, Leopold Aschenbrenner, a former technical program manager at OpenAI, criticized the company’s security measures in a memo to the board. He argued that OpenAI was not doing enough to prevent potential threats from foreign adversaries.
Aschenbrenner was subsequently fired, the newspaper reported.
Aschenbrenner’s concerns were dismissed by OpenAI, which maintained that his departure from the company was unrelated to his security criticisms.
“We appreciate the concerns Leopold raised while at OpenAI, and this did not lead to his separation,” OpenAI spokesperson Liz Bourgeois was quoted as saying to The New York Times. “While we share his commitment to building safe AGI, 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.”
OpenAI did not immediately respond to a request for comment.
Read more:
OpenAI failed to report a major data breach in 2023
A hacker infiltrated OpenAI’s internal messaging system, gaining access to employee discussions regarding the company’s latest AI advancements, according to a New York Times report.
www.csoonline.com
