Two recent federal rulings highlight serious risks of sharing confidential information with public AI platforms. Both decisions confirm what many IP practitioners have long anticipated: users voluntarily inputting proprietary information into consumer-grade generative AI tools can forfeit trade secret protection and undermine confidentiality.

In Trinidad v. OpenAI (N.D. Cal., No. 25-cv-6328, Jan 5, 2026), the court dismissed DTSA claims by a pro se plaintiff who developed her alleged proprietary AI frameworks using consumer ChatGPT. By sharing the information directly with OpenAI absent any confidentiality obligation, she failed the “reasonable measures” requirement under the DTSA. The court applied the Supreme Court’s principle from Ruckelshaus v. Monsanto Co., 467 U.S. 986, 1002 (1984): disclosing a trade secret to others under no duty of secrecy extinguishes the property right.

In United States v. Heppner (S.D.N.Y., No. 25-cr-503, Feb 17, 2026), the court held that documents a criminal defendant generated using Anthropic’s consumer-tier Claude were not protected by attorney-client privilege or work product doctrine. The defendant used the tool independently of counsel, and Anthropic’s privacy policy permitted inputs to be used for training and disclosed to third parties—destroying any reasonable expectation of confidentiality.

Key Distinction for Practitioners

Both cases involved independent use of consumer platforms without attorney involvement. When attorneys direct the use of commercial AI tools featuring data retention restrictions, data processing agreements, and firm-level governance, the risk profile drops materially. Attorney-supervised AI use can help preserve trade secret status and privilege, including via the Kovel doctrine (when the AI functions as a supervised extension of counsel).

Why This Matters

  • DTSA “reasonable measures” should account for AI usage policies and data flows.
  • “Shadow AI”—employees or parties using personal consumer accounts—creates exposure without contractual safeguards.
  • Enterprise licenses mitigate risk, but courts may scrutinize internal policies, employee training, and access controls.

As Peter J. Toren observes in his recent IPWatchdog, Inc analysis, courts are applying longstanding doctrine to AI without carving out special exemptions. The framework exists, and diligence must match the AI era.

Companies with valuable IP portfolios should audit AI usage protocols now. Proactive governance today prevents costly disputes tomorrow.

More background: https://lnkd.in/e3SNuxkA