Skip to content
Vulnerability in OpenAI Skill Exposes Tokens to Unregistered Domain

Vulnerability in OpenAI Skill Exposes Tokens to Unregistered Domain

First seen 17 Oct 2026, 16:47 UTC • •
API

Article Content

Browse articles
ThreatCluster AI
ThreatCluster •October 17, 2026 at 18:49 UTC
  • •A vulnerability in an OpenAI skill allowed tokens to be sent to an unregistered domain.
  • •The flaw was discovered by AIR Security, which registered the domain to demonstrate the risk.
  • •Users are advised to update to the patched version of the skill to prevent token exposure.

Researchers identified a vulnerability in an official OpenAI skill that directed applications to send tokens to an unregistered domain, render-internal.com. This flaw was discovered by AIR Security, which registered the domain and began receiving tokens and API calls from production apps. OpenAI's GitHub repository of skills is generally trusted, but this incident highlights that even vetted skills can contain vulnerabilities. The issue was introduced in a commit made in January 2026 and was resolved by mid-August 2026 when Render removed references to the domain and OpenAI purged the Render plugin. Users are advised to update to the patched version of the skill to mitigate risks. The incident exemplifies the need for ongoing monitoring of AI skills as part of the software supply chain.

Start a free Starter trial for enhanced analysis

Ask AI about this cluster

Updated just now How this analysis works

Timeline

2026-01-01
Vulnerability introduced
A commit in January 2026 introduced the flaw in the OpenAI skill that directed tokens to an unregistered domain.
Csoonline
2026-08-15
Issue resolved
OpenAI and Render resolved the vulnerability by removing references to the unregistered domain from the skill.
Csoonline
2026-10-16
Public disclosure
AIR Security published details of the vulnerability and its implications for AI supply chain security.
air.security

More articles in this cluster (2)

Following this threat?

Track OpenAI in your own feed — you're alerted when they show up in new reporting, leak sites or exploitation.

Free account · no card needed

Extracted Entities (12)

1 / 3

Common questions

What specific skill is affected?
The affected skill is designed for deploying applications on Render and is part of OpenAI's official GitHub repository.
How can I protect my applications?
Update to the patched version of the OpenAI skill to ensure tokens are not sent to the unregistered domain.
What was the nature of the vulnerability?
The vulnerability involved directing tokens to an unregistered domain, which could be exploited by attackers.