Researchers used Claude to hack OpenAI

A security research team used Anthropic’s Claude tooling to breach an OpenAI employee account, exposing internal code and underscoring how AI platforms, third-party services, and identity controls can become high-value attack paths. For CIOs and technology leaders, the incident reinforces that AI adoption expands the enterprise threat surface and that governance, access management, and supply-chain security must be built into AI programs from the start. The article also highlights a broader strategic shift: AI is increasingly being used to build AI, which raises both productivity potential and oversight risks for IT organizations.