Hackers Exploit OpenAI Vulnerabilities Using Anthropic's Claude
Article Content
- •Hacktron exploited OpenAI's forum vulnerabilities using Anthropic's Claude AI.
- •The attack involved a HEIF image upload and a single sign-on flaw.
- •OpenAI fixed the vulnerabilities within 14 hours and paid a $6,500 bounty.
In July 2026, researchers from Hacktron used Anthropic's Claude AI model to exploit vulnerabilities in OpenAI's community forum, leading to unauthorized access to internal repositories. The attack involved a remote code execution flaw in the Discourse forum software, triggered by a malicious HEIF image upload, and a subsequent weakness in OpenAI's single sign-on system. Hacktron demonstrated access by creating a harmless pull request in OpenAI's private GitHub repository using a compromised employee's Codex account. OpenAI responded quickly, fixing the vulnerabilities and issuing a $6,500 bounty for the reported flaws. The incident highlights the risks posed by advanced AI models in cybersecurity. The vulnerabilities were tracked under CVE-2026-32882, which received a CVSS score of 8.8. The attack chain underscores the importance of robust security measures across interconnected systems.
Ask AI about this cluster
Answers cite the sources they use
Timeline
More articles in this cluster (9)
Following this threat?
Track OpenAI and CVE-2026-32882 in your own feed — you're alerted when they show up in new reporting, leak sites or exploitation.
Free account · no card needed