OpenAI Replaces macOS Certificates After Security Incident Linked to Axios Library

OpenAI has decided to replace its macOS certificates following a security incident related to the open-source Axios library, which was involved in a software supply chain attack. Although there is no evidence that user systems or data were compromised, the company has chosen to take preventive measures to mitigate any potential risk.

The vulnerability was discovered on March 31, 2026, when a malicious version of Axios was distributed through an automated process based on GitHub Actions. This system was used by OpenAI to digitally sign its macOS applications, utilizing certificates that ensure the authenticity of the software.

According to OpenAI, it is unlikely that the certificates were actually compromised, thanks to specific technical factors related to the way they were uploaded and the timing of the process. However, the company has decided to treat the incident as if a compromise were possible by revoking the old certificates and introducing new ones.

Further Details on the OpenAI Incident

The main risk, in the case of certificate theft, would have been to allow malicious actors to sign harmful software, making it appear legitimate. To avoid this scenario, OpenAI has released new versions of its macOS applications and collaborated with Apple to prevent the approval of software signed with the previous certificates. Users are therefore encouraged to update OpenAI applications through official channels.

Older versions will no longer receive updates starting from May 8, 2026 and may stop functioning. Investigations have not revealed unauthorized access to user data or manipulations of OpenAI products. Furthermore, there is no evidence that the certificates were actually exploited by attackers. The company has nonetheless commissioned an external forensics firm for an in-depth analysis and conducted additional checks on all signed software.

At the root of the problem was a configuration error in the automated process: the use of a "floating tag" instead of a specific version allowed for the upload of the compromised library. This mechanism has been corrected to prevent future incidents. According to BleepingComputer, the attack is linked to a North Korean group that utilized social engineering techniques to compromise a developer account.