Researchers Hack OpenAI Accounts Using Claude AI
Three independent security researchers claim they breached OpenAI employee accounts in under 72 hours using Anthropic's Claude AI. The breach reportedly gave them access to OpenAI's code repository.
A team of three independent security researchers stated they successfully accessed OpenAI employee accounts using Anthropic's Claude AI model within 72 hours. The group, operating under the name Hacktron, gained entry through OpenAI's GitHub repository, known as "Monorepo."
According to sources cited by The Wall Street Journal, the "Monorepo" contains what are described as "OpenAI's algorithmic secrets." While the researchers did not extract internal code, they demonstrated their access by submitting a pull request from an employee's Codex account.
The entry point for the breach was identified as Discourse, a third-party platform used by OpenAI for its community forums. The researchers leveraged the Claude AI to identify and exploit vulnerabilities within the system.
This incident highlights potential security risks associated with the use of advanced AI models and the protection of sensitive corporate data. OpenAI has not yet issued a public statement regarding the reported breach.