GitHub AI agent leaks private repositories via prompt injection attack

Security researchers at Noma Security discovered a prompt injection attack called GitLost that can trick GitHub's preview Agentic Workflows into leaking private repository contents publicly. An unauthenticated attacker submits a crafted GitHub Issue to a public repo; if the AI agent has read access to private repositories in the same organization, it retrieves and posts sensitive data in a public comment. The attack bypassed GitHub's prompt-based guardrails with minor wording changes and required no stolen credentials or malware. Experts highlight a deeper architectural problem: AI agents operate on broad service account permissions rather than user-scoped permissions, creating invisible trust boundary gaps. Recommended mitigations include explicit repository whitelists, treating all user inputs as untrusted before reaching the LLM, and having an emergency kill-switch for rogue agents.

Nguồn: https://www.infoworld.com/article/4194468/github-ai-agent-leaks-private-repositories-via-prompt-injection-attack-2.html. 8sync News chỉ tóm tắt và dẫn link; bản quyền nội dung thuộc tác giả và nguồn gốc.

Đề xuất cho bạn

Addy Osmani53 Hot9 phút1 ngày trướcAITop

The Agent-Era Career

Addy Osmani, cựu trưởng nhóm kỹ thuật Google, đưa ra lời khuyên sự nghiệp cho kỷ nguyên AI …

#career#productivity#ai-agents