A hacker reportedly obtained access to the internal messaging systems of OpenAI, stealing confidential information on the design process involved with the creation of the company’s AI products.
As reported by the New York Times, an online forum where employees communicate on the ChatGPT maker’s latest developments was compromised but the tech giant has intimated the malicious actor did not access the main repository systems for storing and building its AI.
Senior executives at OpenAI, which has recently launched CriticGPT, briefed the company board and workers on the breach which is said to have taken place in April 2023 but no details were made public as there was no loss of customer or partner data. The report further indicated that bosses believed this was the work of a private individual with no connections to a nation-state backer, nor was it believed to be a national security threat.
OpenAI also did not inform federal law enforcement agencies regarding the incident.
Potential for national security risk
The company is familiar with a sense of alert after it confirmed in May that it had rumbled five covert influence operations that aimed to use its AI models for “deceptive activity” across the internet.
Some employees have expressed fears relating to this security breach, concerned that it could lead to geopolitical rivals like China stealing the firm’s AI property.
Although the technology is primarily a tool for work and search, it is evolving at a rapid pace and eventually, that could lead to a national security risk. Questions have been asked on how serious OpenAI was taking security matters with the findings highlighting flaws within the company on the risks posed by AI.
Leopold Aschenbrenner, a technical manager working on a brief to safeguard the future of AI tech contacted OpenAI’s board of directors to warn them not enough was being done to prevent hostile foreign actors from penetrating its systems.
Aschenbrenner was fired earlier this year for leaking information outside the company but he has insisted his sacking was politically motivated.
The post Hacker stole internal details from OpenAI in 2023 breach, says report appeared first on ReadWrite.