Apple Presents Forensic MacBook Evidence in OpenAI Trade Secret Lawsuit

Apple’s new forensic evidence forces a push for expedited discovery

Apple’s supplemental brief, filed under seal on August 31 2026, contains forensic artifacts from a MacBook taken from former Apple engineer Chang Liu. The brief concludes that the data directly demonstrates Liu’s misuse of Apple’s confidential power‑converter circuit schematic while employed at OpenAI and his intent to delete or hide that evidence.

Key findings from the MacBook forensic dump

Trade‑secret usage – The MacBook logs show Liu opened the Apple circuit file on a Mac mini, which subsequently synced to the MacBook via iCloud. Apple alleges Liu then fed the schematic into an AI‑driven workflow (LTspice simulations) while at OpenAI.
AI‑agent involvement – Internal emails quoted in the brief reveal Liu told a colleague that an AI “agent learned to run LTspice and look at results, tune compensation parameter,” and that the agent performed the simulation in roughly half the time of a manual run.
Evidence‑destruction plan – After Liu learned of Apple’s internal investigation in June, he and a co‑defendant discussed “restoring” and then “starting to use” Apple‑owned devices, language Apple interprets as a plan to wipe logs, metadata, and usage records.
Unexamined devices – The brief notes that a second device, a Mac mini, remains uninvestigated, raising the risk that additional trade‑secret material is still hidden.

"The new evidence confirms Mr. Liu used the trade secrets he misappropriated from Apple while employed by OpenAI. It further confirms that additional pertinent devices remain uninvestigated, that the full scope of Defendants’ misappropriation is yet to be exposed, and that relevant evidence is at risk of degradation and destruction." – Apple’s supplemental brief, §IV.

Why Apple is seeking expedited discovery

Apple argues that the combination of (1) active use of stolen trade secrets in AI‑driven simulations and (2) the defendants’ apparent plan to destroy forensic artifacts creates a unique risk of irreversible harm. Courts have previously granted expedited discovery when evidence is at risk of deletion, and Apple cites recent precedents (e.g., AssuredPartners, 2026 WL 1182808; WeRide, 379 F. Supp. 3d at 854).

Community reaction on Hacker News

dgellow summarized the core allegations: Liu downloaded a confidential Apple schematic, used it at OpenAI, was aware of the unauthorized cloud access, and instructed a colleague to destroy evidence.
joshka highlighted the legal novelty of Apple’s claim that feeding trade secrets into an AI model creates “irreversible and continually propagating uses” of the secret.
biggc raised privacy concerns about iCloud syncing, noting that corporate‑owned devices can expose personal data when they automatically sync to company‑controlled accounts.
pluc and irthomasthomas provided excerpts from Liu’s messages, confirming he boasted about an AI agent that could run LTspice simulations and that he was “feeling AI all day long.”
lrvick warned that once trade secrets are baked into a model, they become practically indefensible, emphasizing the broader industry impact.

Legal and technical implications

Trade‑secret protection vs. AI training – Apple’s argument pushes courts to consider whether a model that has been trained on proprietary schematics can be forced to “unlearn” that information, a question that could set precedent for future IP litigation involving generative AI.
Forensic preservation – The case underscores the importance of immediate forensic imaging of seized devices. Metadata and logs are volatile; delayed analysis can render critical evidence unrecoverable.
Corporate device policies – Companies may need stricter controls over iCloud and other auto‑sync services on employee‑issued hardware to prevent inadvertent leakage of confidential data.

What to watch next

Court ruling on expedited discovery – The judge’s decision on Apple’s motion will determine whether the parties must produce the Mac mini and any additional devices within a tight deadline.
Potential injunctions – If the court finds the AI‑trained models contain Apple’s trade secrets, it could order OpenAI to purge or isolate that data.
Broader industry fallout – A ruling favoring Apple could compel other tech firms to audit their AI training pipelines for inadvertent ingestion of competitor IP.

All quotations and factual statements are drawn directly from the sealed Apple supplemental brief (Case 5:26‑cv‑07078‑EJD, Document 94‑1) and the top‑voted Hacker News comments linked above. No additional information has been inferred or fabricated.