AI Notetaker Lawsuit: Otter.ai Wiretap Claims Move Forward

A ruling that changes the default assumption

AI meeting assistants have spread quickly through corporate calendars. A bot joins the call, listens, and produces a transcript and summary afterward. Many organizations treat that as a routine productivity feature. A federal court in California has now signaled that, at least at the pleading stage, the law may view it very differently.

In August, a federal judge in the Northern District of California denied Otter.ai's motion to dismiss the central wiretap and biometric privacy claims in In re Otter.ai Privacy Litigation. 2 The decision lets a class action proceed against the company over its AI-powered Notetaker meeting assistant. 12

The ruling does not find Otter.ai liable. A denied motion to dismiss means only that the plaintiffs' allegations, taken as true for now, state a viable legal claim. But the court's reasoning goes to the heart of how these tools are built and sold. That is why it matters well beyond one vendor.

The "invited guest" defense fell short

Otter.ai's core argument was intuitive. Notetaker enters a meeting only when invited, the company said, and it functions as the host's own recording tool, an extension of the customer who deployed it. 2 Under that theory, the software is no more an eavesdropper than a tape recorder on the host's desk.

Judge Eumi K. Lee rejected that framing. 2 The plaintiffs allege that Otter.ai does more than hand recordings back to its users. They say it also keeps the captured conversations and voiceprints and uses them to train its own models. 2 Given those allegations, the court found the plaintiffs had adequately claimed that Otter.ai acted as a third-party eavesdropper rather than as a mere tool of the meeting host. 2

The distinction is the key point. When a vendor gains its own benefit from captured data, such as improving its AI, it starts to look less like the customer's instrument and more like an independent party listening in. The participants who never agreed to that party being present are the ones bringing the claims.

Why compliance teams should care

The case is being framed as a warning for organizations, not just AI vendors. Patrick E. Zeller, general counsel of JetStream Security, argues that the companies most likely to suffer from meeting-assistant liabilities are the ones that never stop to assess the risk. 2 Both versions of the coverage cast the dispute in terms of AI compliance exposure. 12

Several practical concerns follow, offered here as analysis rather than holdings from the case:

• Consent is not one-dimensional. A host inviting a bot does not necessarily mean every participant has agreed to recording. It may also not cover a vendor's secondary use of the audio. Internal staff, clients, candidates, and outside counsel may all join calls where a notetaker is quietly running.
• Biometric data raises the stakes. The claims involve voiceprints as well as words. 2 That pushes these tools beyond ordinary recording concerns and into biometric privacy, where obligations and penalties can be stricter.
• Vendor terms matter more than features. Whether a provider retains recordings or trains on them was decisive to the court's reasoning. 2 Procurement and legal teams therefore need to read data-use clauses closely. Settings that disable training or retention may carry real legal weight.

The discovery problem hiding in the transcript

The coverage also flags "discovery headaches," and that concern stands somewhat apart from the wiretap question. 12 Any tool that automatically produces searchable, timestamped records of internal discussions creates a large new body of potential evidence. Offhand remarks that once disappeared when a call ended may now persist in a transcript or an AI-written summary. Those summaries can also contain errors or lack context.

For organizations, this means meeting-assistant output should probably fall under the same retention schedules, legal-hold procedures, and privilege reviews as email. Letting bots record by default, with no governance, risks building an archive nobody designed and few can fully account for.

The takeaway

It is too early to know how In re Otter.ai Privacy Litigation will be resolved. Still, the court's refusal to accept the "extension of the host" theory weakens a comfortable assumption many businesses have relied on. That assumption holds that an invited AI assistant is legally invisible. The safer reading is that these tools bring a vendor into the room. Organizations should decide deliberately whether, when, and on what terms that vendor gets to listen.