NewsResearch
Researchers chain an SSO misconfiguration and a libheif bug to reach OpenAI employee accounts
As reported in The Hacker News weekly recap, researchers used Claude to help chain a misconfiguration in OpenAI's single sign-on with a libheif remote code execution flaw (CVE-2026-32882) in a Discourse community forum, gaining access to employee ChatGPT accounts and internal repositories. OpenAI fixed it within 14 hours of disclosure.
Why it matters
Ordinary web bugs plus an AI assistant that speeds up the chaining. Patch cadence matters more when the attacker works faster.