Overview
Apple has submitted what it describes as ‘shocking evidence’ in its ongoing lawsuit against OpenAI, centred on allegations that former Apple hardware engineer Chang Liu exfiltrated confidential trade secrets for use at his new employer. After Liu’s legal counsel surrendered his old Apple work laptop for forensic investigation, Apple claims it found evidence that Liu used a confidential Apple circuit schematic in his OpenAI work and deployed a tool sharing the name of an internal Apple engineering application. Apple further alleges that Liu enlisted OpenAI colleague Yu-Ting Peng to help destroy evidence once he learned he was under investigation. The case carries significant implications for how AI companies manage sensitive technical IP and enforce offboarding procedures.
Technical Analysis
The case surfaces two technically distinct failure modes. First, Apple alleges Liu exploited a ‘rare, previously unknown authentication bug’ to maintain access to Apple internal systems after his departure — a form of unauthorised persistent access that circumvented standard offboarding controls. This is consistent with patterns where legacy session tokens, misconfigured SSO policies, or unrevoked service account credentials enable continued access well after employment termination.
OpenAI has countered that Liu’s residual access reflects a systemic failure by Apple to properly manage access revocation when employees leave — a common organisational weakness. Regardless of attribution, the authentication vulnerability enabled post-termination data access, which Apple argues was then deliberately weaponised.
Second, the alleged use of a confidential circuit schematic in live OpenAI engineering work represents a direct supply chain integrity risk: proprietary hardware IP from one organisation being incorporated into a competitor’s development pipeline without authorisation.
Framework Mapping
MITRE ATLAS:
- AML.T0012 – Valid Accounts: Liu allegedly maintained authenticated access to Apple systems post-departure via an exploited authentication bug, mirroring the use of legitimate credentials for unauthorised access.
- AML.T0044 – Full AI Model Access: The alleged use of Apple’s internal engineering tools and schematics at OpenAI suggests broad access to proprietary AI hardware design assets.
- AML.T0010 – AI Supply Chain Compromise: If Apple’s hardware IP was integrated into OpenAI’s development pipeline, this constitutes a supply chain integrity breach from Apple’s perspective.
OWASP LLM Top 10:
- LLM10 – Model Theft: The alleged exfiltration of circuit schematics and engineering tools maps to theft of AI-adjacent infrastructure assets.
- LLM06 – Sensitive Information Disclosure: Confidential technical data was allegedly accessed and transferred outside its intended security boundary.
- LLM05 – Supply Chain Vulnerabilities: Proprietary Apple technology allegedly entering OpenAI’s development environment represents a supply chain integrity failure.
Impact Assessment
The immediate impact is legal and reputational for both Apple and OpenAI. Apple is seeking a preliminary injunction that would halt OpenAI hardware development activities predicated on Apple technology — a potentially significant operational disruption. The broader industry implication is substantial: with over 400 former Apple employees now at OpenAI, Apple’s request for expedited discovery suggests the scope of potential IP exposure may extend far beyond Liu alone. For the AI industry, this case underscores the systemic risk of talent mobility between direct competitors without adequate data hygiene controls.
Mitigation & Recommendations
- Enforce zero-trust offboarding: Revoke all credentials, tokens, and device access simultaneously at the moment of employment termination — do not rely on staggered or self-service processes.
- Audit authentication logs post-departure: Monitor for any access attempts by former employees and investigate unknown authentication pathways immediately.
- Deploy endpoint DLP controls: Prevent bulk file transfers or access to sensitive schematics from personal or transitional devices in the weeks before and after an employee’s departure date.
- Conduct IP exit interviews and forensic baseline checks: Establish a documented baseline of what data a departing employee had access to, and require device return prior to final pay release.
- Patch authentication vulnerabilities proactively: Unknown authentication bugs that persist post-offboarding represent a systemic access control failure that internal red teams and auth audits should surface.