- Third-party tool Axios compromised in wider cyber attack
- No evidence of user data or systems being affected
- Older app versions may stop working after May 8
OpenAI has flagged a security issue affecting its desktop apps, including ChatGPT, after a wider software supply chain attack compromised a commonly used developer tool.
The issue centres on Axios, a behind-the-scenes software library used by developers. It was compromised on March 31 as part of a broader cyber attack reportedly linked to North Korea, raising concerns across the tech industry. OpenAI said the incident briefly affected part of the system it uses to verify that its apps are genuine.
For users, this mainly impacts those using ChatGPT and other OpenAI tools as downloaded apps on Apple laptops and computers.
The company said there is no evidence that user data was accessed, accounts were breached, or its systems were compromised.
Where things went wrong
According to OpenAI, the issue began when an internal automated process — known as a GitHub workflow — unknowingly downloaded a malicious version of Axios.
This process had access to sensitive app-signing infrastructure, including digital certificates used to confirm that apps like ChatGPT Desktop, Codex and Atlas are officially from OpenAI.
These certificates act like a trust badge. If misused, they could allow fake apps to appear legitimate.
OpenAI said its analysis suggests the certificate was likely not stolen, reportedly said in a news report, based on how the attack unfolded and existing safeguards. Still, the company is treating it as potentially compromised.
Precaution, not panic
As a result, OpenAI is revoking and replacing its security certificates and tightening its app verification process.
Users are being asked to update their apps to the latest version. Older versions will stop receiving updates and may stop working after May 8.
The company said this step is meant to prevent even a small chance of fake apps being distributed under its name. It also confirmed that passwords and API keys were not affected.
The episode highlights how supply chain attacks are becoming a weak spot for the tech industry. Instead of targeting companies directly, attackers are going after the tools developers rely on — which can quietly open doors into larger systems, even if only for a short time.
