Industry News

AI Tool Reportedly Used to Access OpenAI's Internal Code

WSJ · 18 Sept 2026
Key Takeaway Small businesses should assume AI tools can now speed up both legitimate security testing and malicious hacking, so keep code repositories, credentials, and internal systems tightly access-controlled and monitored.

Reports indicate that an independent security research team used Anthropic's Claude AI system to gain access to OpenAI's internal code repositories. Details on the exact method and scope of access remain limited.

The incident points to a broader trend of AI tools being used to assist in both offensive and defensive cybersecurity work, including bug hunting. As AI systems become more capable, they may increasingly be used by researchers and attackers alike to probe software systems for weaknesses.

Building or buying AI systems? Governing them under ISO 42001 ->

Summarised by CISO AI from WSJ. We link back to every original so you can read it yourself.