MalwareBleeping Computer
8.0 — CRITICAL
'Ghostcommit' hides prompt injection in images to fool AI agents, steal secrets
A PNG hiding a prompt injection could steal your repo's secrets, researchers demonstrate. The technique, dubbed 'Ghostcommit,' slipped past AI code reviewers CodeRabbit and Bugbot, which never open image files at all, then convinced a coding agent to read a repo's .env and write every secret into the code as a list of numbers. [...]
🤖 AI BriefingAuto-generated threat analysis
🔍Threat Overview
The 'Ghostcommit' attack exploits a review gap in AI code reviewers, hiding malicious instructions inside images that are never opened by human reviewers, allowing attackers to steal repository secrets. The attack relies on the tooling deciding what to do with the image rather than the model.
⚙️Technical Details
💥Impact Assessment
Severity: High
🛡️Recommended Actions
1Implement multimodal pull-request defenders to scan for invisible characters, code shape, and images
2Monitor agent behavior when reading credentials files
3Regularly review and update AGENTS.md convention files for suspicious activity
📦Affected Products
Software:GitHubGitLab
Read the full article
This is a curated summary. The complete article is available at Bleeping Computer.