# CVE-2026-33232: Significant-Gravitas AutoGPT Denial of Service via disk exhaustion

Source: https://junglewise.ai/threats/cve-2026-33232-significant-gravitas-autogpt-denial-of-service-via-disk  
Publisher: Junglewise Threat Intelligence, https://junglewise.ai/threats  
Published: 2026-05-19. Updated: 2026-05-19  
License: CC BY 4.0 with a link back

CVE-2026-33232 · Severity: high · CVSS 7.5 · Published 2026-05-19

AutoGPT is a platform used to create and manage autonomous AI agents. A security flaw in its backend allows anyone on the internet to crash the service by repeatedly requesting a specific file download. This process fills up the server's hard drive with temporary files that are never deleted, eventually causing the database and the entire platform to stop working for all users.

Technologies: [Significant-Gravitas AutoGPT Platform Backend](https://junglewise.ai/threats/technologies/autogpt-platform-backend). Vendors: [Significant-Gravitas](https://junglewise.ai/threats/vendors/significant-gravitas).

## Quick answers

**Is CVE-2026-33232 exploited in the wild?** No. Junglewise Threat Intelligence has no report of CVE-2026-33232 being exploited in the wild.

**How severe is CVE-2026-33232?** High severity, with a CVSS score of 7.5, published 19 May 2026.

**Which products and versions are affected?** Significant-Gravitas AutoGPT Platform Backend 0.4.2 - 0.6.51.

**Is there a fix for CVE-2026-33232?** A patch was published on 2026-03-20: Version 0.6.52 released with fix Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.

**How is CVE-2026-33232 attacked?** It is attacked over the network, in a network appliance.

**Am I affected by CVE-2026-33232?** Check whether you run Significant-Gravitas AutoGPT Platform Backend at an affected version. Junglewise does that check for you around the clock: connect your code, cloud and domains, and it matches new vulnerabilities like this one against what you actually run and tells you when one reaches you.

## Technical details

A Denial of Service (DoS) vulnerability exists in the AutoGPT platform backend due to improper resource cleanup in the `download_agent_file` endpoint. The vulnerable function uses `tempfile.NamedTemporaryFile` with `delete=False` to generate agent files but fails to implement a mechanism to delete these files after they are served by FastAPI's `FileResponse`. Because this endpoint is publicly accessible and requires no authentication, a remote attacker can automate repeated requests to fill the host's temporary storage directory. This leads to 'No space left on device' errors, causing the backend database and other critical system services to fail. The issue is addressed in version 0.6.52 by ensuring temporary files are properly removed.

## Affected products

- Significant-Gravitas AutoGPT Platform Backend 0.4.2 - 0.6.51

## Timeline

- 2026-05-11: advisory: GitHub Security Advisory GHSA-374w-2pxq-c9jp published
- 2026-05-18: disclosed: CVE-2026-33232 published to NVD
- 2026-03-20: patched: Version 0.6.52 released with fix

## References

- [https://github.com/Significant-Gravitas/AutoGPT/releases/tag/autogpt-platform-beta-v0.6.52](https://github.com/Significant-Gravitas/AutoGPT/releases/tag/autogpt-platform-beta-v0.6.52)
- [https://github.com/Significant-Gravitas/AutoGPT/security/advisories/GHSA-374w-2pxq-c9jp](https://github.com/Significant-Gravitas/AutoGPT/security/advisories/GHSA-374w-2pxq-c9jp)

## Related threats

- [Significant-Gravitas AutoGPT SSRF in SendEmailBlock](https://junglewise.ai/threats/cve-2026-33234-significant-gravitas-autogpt-ssrf-in-sendemailblock)
- [Significant-Gravitas AutoGPT unsafe deserialization in Redis cache](https://junglewise.ai/threats/cve-2026-33233-significant-gravitas-autogpt-unsafe-deserialization-in-redis)
- [Significant-Gravitas AutoGPT session hijacking via IDOR](https://junglewise.ai/threats/cve-2026-30950-significant-gravitas-autogpt-session-hijacking-via-idor)