Researchers used Claude to hack OpenAI
Researchers used Claude to chain a libheif vulnerability in Discourse with an OpenAI SSO flaw, gaining access to an internal GitHub repository.
Researchers at Hacktron used Anthropic's Claude to discover and chain two vulnerabilities affecting OpenAI: a libheif flaw in Discourse's image processing and an SSO vulnerability. This allowed access to an internal OpenAI GitHub repository via an employee's Codex account. The researchers reported the flaw, received a $6,500 bounty, and discovered related issues in Slack, Meta, and GitHub as part of a larger 'HEIF Heist' project.
- Hacktron used Anthropic's Claude to chain a libheif vulnerability in Discourse with an OpenAI SSO flaw.
- The exploit granted access to OpenAI's internal GitHub repository via an employee's Codex account.
- OpenAI paid a $6,500 bounty; related flaws were found in Slack, Meta, and GitHub.
- Vendors
- Malwarebytes
From the original article
4 min read at malwarebytes.com
Claude helped researchers break into OpenAI in under 72 hours, and exposed how quickly AI is lowering the bar for sophisticated hacking.
By @malwarebytes · malwarebytes.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.