A critical vulnerability in the Paperclip AI management platform, tracked as CVE-2026-41679, allowed attackers to bypass authorization and execute arbitrary code with server permissions.

Key Takeaways

  • A critical vulnerability (CVE-2026-41679) with a CVSS score of 10 was discovered in Paperclip.
  • Attackers could self-register and gain board-level API access without email verification.
  • The flaw enabled remote code execution (RCE) via malicious company imports.
  • Oasis Security identified the flaw, and patches have been deployed to fix the issue.

A massive authorization bypass vulnerability has been identified in Paperclip, an AI management platform designed to scale autonomous AI agents. According to a report by Oasis Security, this flaw could have allowed remote attackers to gain arbitrary code execution with the full permissions of the server.

Technical Breakdown of the Exploit

Tracked as CVE-2026-41679, the vulnerability carries a maximum CVSS score of 10. The exploit path was alarmingly simple: an attacker could self-register an account without needing email verification, sign in, and manipulate the CLI authorization flow. By self-approving a challenge, the attacker could obtain a board-level API token, which provided sufficient access to reach company import routes.

Why This Matters

BozokMedia analysis shows that this isn't just a simple bug; it is a fundamental failure in how authorization was scoped. While creating a new company required instance-admin privileges, the import path only required board-level access. This discrepancy allowed attackers to use a crafted .paperclip.yaml file to execute commands as the Paperclip server process, potentially exposing application data, source repositories, and local secrets.

As AI agents become a new class of enterprise identity, the distributed nature of agentic workflows creates unprecedented security blind spots.

In addition to the main flaw, Paperclip has also patched two other vulnerabilities: one involving sensitive data disclosure via API routes and another involving DNS rebinding that could target developer machines in local-development mode.

Frequently Asked Questions

1. How severe is this vulnerability?
It is rated as 'Critical' (CVSS 10), meaning it poses the highest possible risk to system integrity.

2. Is there a fix available?
Yes, Paperclip has released updates that implement stricter authorization checks and tighten company scoping.

Did You Know?: A CVSS score of 10 is the highest possible rating, indicating a vulnerability that is easy to exploit and has a devastating impact.