OpenAI publishes messages and emails against Apple’s trade-secrets suit

In Apple is getting this wrong, OpenAI published iMessages and legal correspondence and denies holding or wanting Apple trade secrets. It is a sharp public counter to the July lawsuit and to Apple’s push for a preliminary injunction around people who left Apple hardware for OpenAI.

OpenAI put iMessages and emails on the table, not only a press denial

The 3 August 2026 post calls the suit “careless, aggressive and oddly personal.” OpenAI says Apple had claimed a February outreach with no reply, but outside counsel emailed the wrong person after confusing two Asian last names. Apple also claimed a discussion with OpenAI’s general counsel that OpenAI says never happened. The company adds that the specific lawsuit allegations were not raised then, and that after a note about “resolving any issues” five months of silence followed until the complaint landed.

On Chang Liu (last Apple day 22 January 2026) OpenAI publishes threads where former colleagues ask him to help locate files and information for Apple work. OpenAI frames that as residual access from poor offboarding controls, not a theft program for the new employer. On Tang Tan (more than 24 years at Apple, now in a hardware role at OpenAI) the company says the team was told not to want or use other firms’ confidential information. OpenAI calls the preliminary-injunction request unnecessary: “we do not have, nor want, any of their trade secrets.”

The court docket is becoming a PR fight over talent and AI hardware

In the same wave Apple seeks a preliminary injunction and expedited discovery. Public reports name Liu and Tang Yew Tan, mention additional former employees, and point to a hearing around 1 October 2026. The fight sits where labs compete for people who know phones, displays and power systems, because OpenAI (and the device-startup orbit around it) is pushing into hardware.

For legal, security and HR teams the practical point is blunt. Offboarding, cutting iCloud and internal drive access, and rules for ex-employee contact with the old team stop being back-office chores. They become exhibit material in a media fight. OpenAI bets readers will see messy departure help-seeking more than a clean exfiltration plan. Apple bets the court will see a pattern around sensitive files, suppliers and recruiting.

Public screenshots are not a judicial finding

What OpenAI published is a one-sided selection. Editors see redacted iMessages and an email trail about a misaddressed legal note. They do not see disk forensics, a full file list or Apple’s claimed chain about repeated downloads. Apple’s filings about more people and imminent trade-secret harm are still procedural moves, not a final judgment.

The second risk is narrative shortcut. “Confused Asian surnames” and “a colleague asked for AirDrop” are strong media images. They can eclipse harder legal questions: what an ex-employee may still know, what may be said in recruiting, and how residual knowledge differs from residual access to files.

Watch the October hearing and what discovery still supports

The blog tone will not decide this. Three signals will. First, whether the court grants a preliminary injunction and how far it constrains the named people and OpenAI’s hardware branch. Second, what remains of claims about additional ex-employees and specific documents after expedited discovery. Third, whether both sides return to a quiet settlement or keep a public evidence ping-pong through the merits.

While the media war runs beside the San Jose docket, treat every “smoking gun” screenshot as a party’s opening position, not a closed forensic finding.

Lilith's verdict

OpenAI is not trying the case in San Jose first. It is trying it in timelines: colleague AirDrops and a misaddressed legal email are meant to outrun Apple’s theft narrative around hardware secrets.