AI Meeting Assistants Go Mainstream as Wiretap Lawsuits Mount

A crowded category reaches its awkward adolescence

In February, Reclaim.ai published a list of the 18 best AI meeting assistants of 2026. Its top pick was Zoom AI Companion for in-meeting summaries, and it put its own product second as the best overall choice. The rest of the list covered sales-intelligence platforms like Gong and Avoma, transcript-first tools like Otter.ai and Fireflies.ai, bot-free apps like Krisp and Granola, and calendar products such as Calendly, whose notetaker was still waitlist-only at the time.5 Eight months later, that list looks less like a buyer's guide and more like a record of a category growing past the novelty stage. The tools are now everywhere, nearly all of them do the basic job well enough, and the open questions have moved from "which transcript is cleanest" to "who consented, who owns the recording, and whose job does this change."

The 2026 roundups show an odd mix of crowding and agreement. Nearly every comparison lists the same core group: Otter, Fireflies, Fathom, Granola, tl;dv, Fellow and the notetakers built into the big suites.189 Who comes out on top mostly depends on who wrote the list. Fellow names Fellow the best assistant for most teams.9 tl;dv puts tl;dv first.6 Laxis says it won after testing five apps in more than 200 meetings.3 Hedy calls itself the best fit for real-time coaching.7 Lindy ranks Lindy first for turning notes into tasks.10 TechRepublic, which doesn't sell a competing product, picked Fathom as best overall because of its generous free plan and its option to record with or without a bot. It gave Fireflies the high-volume teams slot, Otter live transcription and Granola bot-free capture.8 Readers should treat most "best of 2026" lists as marketing, and look for the places where rival vendors agree with each other.

Where the reviews agree, and where they don't

One point of agreement is that transcription is no longer what sets products apart. Simular's hands-on review put it most bluntly: most tools stop at summaries, and the real contest is over what happens after the call.1 Cirrus Insight's comparison is built around CRM automation before, during and after meetings.2 Reclaim praised Avoma for pushing work forward rather than just storing transcripts.5 Fellow's guide now checks whether each tool exposes meeting data to AI agents through MCP servers. It notes that Fathom offers official Claude and ChatGPT connectors on every plan, while Otter's connector supports search but no documented write actions.9 The direction is clear. Meeting assistants are turning into workflow automation layers that feed CRMs, task lists and AI agents, and the meeting record is becoming raw material rather than the product.

Pricing is where the coverage disagrees most, and much of the gap comes from how plans are billed. Simular lists Otter's paid tier at $16.99 a month.1 TechRepublic and Fellow quote $8.33 a month billed annually.89 Fireflies shows up at $18 in one comparison and $10 on annual billing in others.18 The free tiers are less generous than they look. Laxis found that Fathom's unlimited recording comes with only five AI summaries a month, Otter cuts free transcripts off at 30 minutes per conversation, and Granola's free plan keeps only a limited note history.3 Buyers should compare the limits, not the headline prices.

The reviews also disagree about bots. Simular framed bot-free tools like Granola, Fathom and Krisp as the better choice for client calls and HR conversations. It also said, approvingly, that Granola's clients never know they're being recorded.1 A courtroom is now testing that claim.

Adoption numbers don't agree with each other

The number most often cited for adoption is 75%: three in four professionals using an AI note-taker in work meetings, with about two-thirds of Fortune 500 companies running one somewhere.2125 It comes from a Fellow.ai survey, and at least one statistics roundup flags it as vendor-funded. That same roundup notes that Federal Reserve data puts overall business AI adoption closer to 18% to 20% of US firms as of late 2025.22 Independent surveys come in much lower. A July 2026 Pollfish survey of 500 employed Americans, commissioned by Kolmogorov Law, found that 33.4% had seen an AI notetaker in their meetings, and another 22.4% couldn't say whether they had.23 A separate survey of 1,000 full-time workers found that just 19% use AI for meeting notes frequently, while 40% never do.27

These figures can all be roughly true at once, because they measure different things: exposure, occasional use and habitual use. The fairest reading is that the tools are widespread but shallowly used. Adoption also runs backwards by company size. One dataset puts it at 74% in firms with 201–1,000 employees and 43% in firms above 5,000, where security reviews and shadow-IT worries slow rollouts.28 Most of the growth comes from individuals bringing their own tools rather than from top-down purchases, which matters for the legal story below.

The productivity claims need the same caution. Vendor-compiled figures say 62% of users save about four hours a week, and the same data says 84% of people change what they say when an AI is listening.2122 One critic points to the gap between 75% adoption and roughly a 25% drop in meeting volume, and argues that notetakers multiply output rather than cut the number of meetings.26 The workforce survey points to a quieter cost: nearly three in ten workers admitted skipping meetings because they trusted the AI summary, and 41% said they had missed context the tool failed to capture.27

The consent reckoning

The legal cases are the biggest development of 2026, and the reporting on them agrees closely. On August 13, Judge Eumi K. Lee of the Northern District of California partly granted Otter's motion to dismiss in the consolidated In re Otter.AI Privacy Litigation. The federal Wiretap Act, California Invasion of Privacy Act and Illinois biometric claims survived and now move to discovery.1119 The key line in the order is the court's finding that the plaintiffs plausibly alleged Otter collects, retains and uses communications for its own commercial purposes, which would make it a third-party eavesdropper rather than a neutral tool of the meeting host.1219 That is a ruling on whether the claims can proceed, not a finding of liability.12

The Granola case shows why removing the visible bot doesn't solve the problem. Chamberlain v. Granola, filed July 30 in the same district, alleges the app recorded a participant who had no notice and used meeting content for model training by default.1117 The complaint reportedly cites Granola's own marketing, which says other participants won't know the app is there, as evidence that the concealment was intentional.1118 Fireflies faces Illinois biometric-privacy suits, and a suit over Microsoft Teams transcription was filed in February.1213 The commercial stakes are high. Granola raised $125 million at a $1.5 billion valuation in March.20 tl;dv calculates that statutory damages could become very large given Otter's reported user base of more than 35 million.14

The consent gap these suits target is not hypothetical. In the Kolmogorov Law survey, 65.3% of workers who had encountered a notetaker said consent was inconsistent or absent, and only about 35% knew that some states require every party's consent.23 Legal commentators broadly agree that the burden still falls mostly on the person who runs the meeting.1119 My reading is that whether a bot joins the call was never the real legal question. Default settings that send recordings into training data and capture that participants can't detect are the exposure, and vendors that make disclosure non-optional will hold up better than those that sell invisibility as a feature.

The labor-market angle: tasks, not yet jobs

Meeting assistants automate a task that has historically belonged to one occupation: taking minutes. In its 2024–34 projections, the Bureau of Labor Statistics expects declines for executive, legal and general secretaries and administrative assistants. It explicitly links limited demand in office support roles to AI-driven efficiency gains.31 Anthropic's own exposure research estimates that LLMs could take on the majority of tasks in office and administrative occupations.36 About 86% of the clerical workers that Brookings identifies as least able to adapt are women.3438

On the ground, the picture is mixed. One Vanderbilt executive assistant told the AP she no longer takes meeting notes because Copilot and ChatGPT handle it, which frees her to take part in the meetings.34 The American Society of Administrative Professionals reports that AI adoption among its members tripled to 77% in two years.38 A laid-off assistant told The 19th that strategic admin roles once paid $70,000–$80,000 now pay six figures, but that employers want one or two of those people where they used to hire four or five.38 That pattern of fewer, better-paid jobs is what you would expect when a tool removes the routine core of a role.

Broader research hasn't settled the question. A recent survey of studies finds aggregate employment stable alongside displacement in specific groups. It cites evidence that AI adoption has reduced junior hiring in some datasets, finds null effects in others, and reports that around 90% of surveyed executives say AI hasn't yet significantly affected employment.33 Meeting assistants fit that pattern. They haven't eliminated jobs on their own, but they are steadily removing the parts of administrative work that were easiest to delegate and hardest to defend.

The bottom line

The 2026 meeting-assistant market is mature enough that features have become commodities and serious enough to draw federal wiretap litigation. The tools that last are likely to be the ones that turn conversations into governed, consented workflow data. The tools that sold on convenience and invisibility face a much harder year.