OpenAI reveals security incident affecting macOS apps after supply chain attack

OpenAI Flags Software Supply Chain Attack on macOS Apps

OpenAI disclosed a security incident on Friday that exposed its macOS apps to a potential software supply chain attack, prompting the company to overhaul its security certification process and mandate immediate user updates 1

1

. The breach occurred on March 31 when a GitHub Actions workflow used to sign certificates for macOS applications downloaded a malicious update from the Axios developer library, a widely used JavaScript library for making HTTP requests that is unaffiliated with Axios Media 1

1

. Hackers who hijacked a developer's account published two infected updates to the Axios library before detection, creating a vulnerability that could have allowed attackers to exfiltrate certificates and create fake OpenAI apps that would appear legitimate to devices and the App Store 1

1

.

North Korean Hackers Linked to Broader Campaign

Google Threat Intelligence Group connected the wider hacking campaign to a North Korean hacker group, underscoring how AI companies have become prime targets for classic software supply chain attacks alongside novel AI-specific threats 1

1

2

2

. The compromise affected the company's GitHub workflow, which could reach signing certificates and notarization materials used to authenticate macOS versions of ChatGPT, Codex, and Atlas 3

3

. Despite the severity of the vulnerability, OpenAI emphasized that there was no evidence any user data, intellectual property, or internal systems were compromised, and no signs that iOS, Android, Windows, or other platforms' apps were affected 1

1

2

2

.

Mandatory Updates and Certificate Revocation by May 8

In response to the security incident, OpenAI is implementing stringent macOS app verification protocols and revoking existing security certifications out of an abundance of caution 2

2

. The company will stop supporting older versions of its macOS apps on May 8, giving users a 30-day window to update before the revoked certificate could block new downloads and first-time launches 1

1

. Mac users of ChatGPT, Codex, and Atlas are required to upgrade to the newest versions to ensure compliance with new security protocols, as older versions will no longer receive updates or support and may become non-functional 2

2

. The root cause was identified as a misconfiguration in the GitHub Actions workflow, which has since been addressed 2

2

.

Implications for Cybersecurity in AI Industry

The incident highlights the escalating cybersecurity challenges facing AI companies as they become high-value targets for state-sponsored actors and sophisticated threat groups. OpenAI confirmed that user passwords and OpenAI API keys were unaffected by the potential breach, and no evidence of malware signed as OpenAI had been detected 2

2

. The company is reportedly finalizing a model with enhanced cybersecurity features through its Trusted Access for Cyber program, which it plans to deploy to a select group of companies, reflecting its commitment to addressing security concerns alongside its growth trajectory 3

3

. This proactive approach to macOS app verification effectively turns patching into a gatekeeper for app legitimacy, aiming to reduce the odds that forged builds can circulate with credible-looking signing 3

3

. The timing is particularly significant as OpenAI navigates broader scrutiny and competition in the AI sector, with the company framing the move as a preventative step rather than a response to confirmed user data theft 3

3

.