Two contradictory rulings issued on April 23, 2026 have thrown the legal status of AI-assisted communications into chaos, with Masimo CEO Joe Kiani's recovered ChatGPT logs now admissible as evidence in court.
If you have ever typed a legal strategy into ChatGPT and assumed it stayed between you and your lawyer, today's events should make you reconsider that assumption entirely. Judge Carl Nichols ruled in the Masimo case that conversations between CEO Joe Kiani and ChatGPT carry no attorney-client privilege whatsoever , and the deleted chat logs Kiani thought were gone were recovered through metadata analysis and admitted into evidence. That is not a hypothetical warning. That is a CEO's private AI sessions sitting in a courtroom record.
The ruling rests on a straightforward but consequential legal logic: privilege exists to protect the relationship between a client and a human attorney. When Kiani used ChatGPT to draft legal strategy, no licensed lawyer reviewed or adopted that output before it shaped his decisions. In the court's view, that makes the AI a tool, not counsel , and tool usage has never been shielded from discovery. The fact that Kiani subsequently deleted the conversations did not help his case. Federal investigators used metadata recovery techniques to reconstruct the logs, establishing not just that the chats existed but that they contained substantive legal deliberation.
Hours later, a separate judge in what analysts are describing as the same circuit issued a ruling that pointed in the opposite direction. That court determined advanced AI models are sufficiently sophisticated to be treated analogously to human legal experts, and extended privilege protections accordingly. The specific case details are still emerging, but the legal effect is immediate: two sitting federal judges now hold irreconcilable positions on the same question, creating a circuit split that neither corporations nor their attorneys can navigate cleanly.
A circuit split of this nature typically means years of uncertainty before an appellate court or the Supreme Court steps in to resolve the conflict. Until that happens, the answer to whether your AI-assisted legal work is discoverable depends heavily on which courthouse your dispute lands in.
What This Means for the Market
The corporate response is likely to be swift and blunt. Legal departments at large firms rarely tolerate ambiguity around discovery risk, and the Kiani case gives general counsel a concrete cautionary tale to cite when pushing for internal AI restrictions. Expect generative AI bans to spread through legal, finance, and executive functions at companies where internal deliberations carry litigation exposure , which is most companies of any significant size.
Legal technology firms and enterprise AI vendors face a more complex problem. The value proposition of AI in legal workflows depends partly on users trusting that sensitive inputs remain protected. The Nichols ruling does not just create liability for one CEO; it signals that any substantive legal reasoning conducted through an AI interface could theoretically surface in opposing counsel's discovery requests. That is a chilling dynamic for a market that has spent the past two years selling AI as a productivity layer for high-stakes professional work.
There is also a data hygiene implication that goes beyond legal strategy. The recovery of Kiani's deleted logs demonstrates that deletion within an AI platform's interface does not constitute the kind of evidence destruction that holds up legally. Organizations that believed they were managing their data footprint by deleting AI sessions now have reason to audit that assumption with their IT and legal teams.
What to watch next is whether OpenAI, Anthropic, and similar providers move to clarify their data retention and recovery policies in response , and whether enterprise legal teams begin demanding contractual guarantees around log persistence. The more immediate question is whether any appellate court fast-tracks a consolidated review of the two rulings, given how quickly the uncertainty could ripple through corporate AI adoption. Until there is a unified standard, the safest working assumption for any executive using an AI tool to think through legal exposure is the one Judge Nichols just established: the conversation may not stay private.
Also read: SK Hynix posts a five-fold profit jump as AI chip demand blows past its ability to supply • Anthropic hits a trillion-dollar valuation on secondary markets and knocks OpenAI from the top spot • OpenAI converts to a public benefit corporation and moves closer to a conventional tech giant
