Our response to the TanStack npm supply chain attack

OpenAI has outlined its response to a security issue involving the TanStack npm library, part of a broader attack known as Mini Shai-Hulud. According to the company, two employee devices in its corporate environment were impacted, resulting in unauthorized access and credential-focused exfiltration within a limited subset of internal source code repositories. OpenAI engaged a third-party digital forensics and incident response firm, isolated impacted systems, revoked user sessions, and rotated credentials. The company confirmed there is no evidence that user data was accessed, that production systems or intellectual property were compromised, or that published software was altered. Because the impacted repositories included code-signing certificates for iOS, macOS, and Windows products, OpenAI is rotating its code-signing certificates as a precaution. While Windows and iOS users do not need to take action, macOS users must update their applications. OpenAI initially noted a June 12, 2026 update deadline, but later extended the timeline to June 26, 2026, after coordinating with Apple to protect users. New notarization with the previous certificate has been blocked, and older versions of macOS desktop apps signed with the outdated certificate may not function or receive support after the transition. OpenAI stated that the incident occurred during the phased rollout of updated security controls, such as hardened CI/CD credentials and package manager configurations. The company emphasizes that attackers are increasingly targeting shared software dependencies, underscoring the need for robust verification of third-party component provenance across the AI industry.