GoKawiilHacktron researchers used Anthropic's Claude to breach OpenAI employee accounts
Three independent security researchers at Hacktron reportedly used Anthropic's Claude Opus models to gain access to OpenAI employee accounts within 72 hours, exploiting a HEIF image-processing flaw in Discourse, the third-party service running OpenAI's community forums. They proved access by submitting a pull request through a compromised employee's Codex account but stopped short of touching OpenAI's proprietary code in its 'Monorepo' repository.
GoKawiil's interpretation of the reporting above, not reported fact.
The exploit shows how quickly AI coding tools can be turned into offensive hacking instruments, drastically lowering the cost and time barrier for finding critical vulnerabilities across major tech platforms. It also raises concerns about supply-chain security, since the weakness lay not in OpenAI's own systems but in a third-party service it relies on, and about how prepared AI labs are against more sophisticated state-backed attackers.
- Hacktron used Claude Opus 4.8 and 5 to breach OpenAI employee accounts via a Discourse vulnerability in under 72 hours.
- The same exploit, called HEIF Heist, was adapted to target Slack, Meta, GitHub, Shopify and others for under $3,000 in AI tokens.
- OpenAI paid Hacktron $6,500 for the bug report, and the researchers noted their methods are far less advanced than nation-state hacking groups.
Source: theverge.com — Stevie Bonifield, 2026-09-18
Published there as: “Security researchers used Claude to help them hack into OpenAI”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.