Hacker stole AI technology details from OpenAI, raising concerns about national security risks.

From Neowin: 2024-07-04 13:16:00

In early 2023, a hacker accessed OpenAI’s internal systems, stealing information on AI technologies. Not disclosed to the public or law enforcement, the cybercriminal breached forums where employees discussed latest technologies but did not infiltrate GPT models’ housing systems.

Concerns emerge that stolen AI technology may be used by countries like China, posing national security risks. Employees question OpenAI’s security measures, leading to internal divisions on AI risks. Former OpenAI manager Leopold Aschenbrenner alleges company negligence in stopping thefts by foreign adversaries, citing a politically motivated dismissal.

The breach at OpenAI exacerbates existing internal issues. CEO Altman’s board disputes, recent researcher departures over superalignment disagreements, and Aschenbrenner’s wrongful termination highlight company instability. AI safety researcher exodus hints at internal turmoil regarding control of superintelligent AI, impacting OpenAI’s future trajectory.

AI experts warn about the potential risks of advanced generative AI technology theft to national security. While current designs being stolen may not pose imminent danger, the evolving capabilities of this technology could increase threat levels over time. Prominent figures emphasize the importance of safeguarding AI technology to prevent potential security breaches.



Read more at Neowin: NYT: Hacker stole details about OpenAI’s AI tech, but the company kept it quiet