CVE-2026-45023 AutoGPT: Credit system bypassed via direct block execution in POST /api/blocks/{block_id}/execute
Attackers can bypass credit limits and execute AI blocks for free by calling the direct execution API endpoint instead of the standard graph execution path.
- Exploited in the wild
- Not known
- Login needed
- Any signed-in user
- Attackable
- Over the internet
- Victim action needed
- Not needed
- Fix
- None listed
What is CVE-2026-45023?
CVE-2026-45023 is a medium-severity vulnerability in Significant-Gravitas AutoGPT, published on 28 May 2026. Any signed-in user can exploit it over the network, with no action from the victim.
Official description
AutoGPT is a workflow automation platform for creating, deploying, and managing continuous artificial intelligence agents. Prior to 0.6.59, POST /api/blocks/{block_id}/execute endpoint executes blocks without consuming any credits, regardless of the user's balance. The credit check that exists in the graph execution path (manager.py) is never reached when blocks are called directly via the external API, allowing unlimited free execution of all blocks. This vulnerability is fixed in 0.6.59.
Which versions are affected by CVE-2026-45023?
How do I fix CVE-2026-45023?
No fix has been published for CVE-2026-45023 yet. Limit exposure of the affected software until one is available.
Is CVE-2026-45023 being exploited?
It is not confirmed as exploited in the wild, but public exploit code for CVE-2026-45023 exists.
Sources
Also known as GHSA-8pjg-mfqm-vrhr.