A proposed class action filed July 30 in the U.S. District Court for the Northern District of California accuses Granola, a venture-backed AI meeting notetaker, of recording conversations without telling most participants and feeding those recordings into its own AI model training by default, according to the complaint.
Chamberlain v. Granola is in the same district as Otter.AI Privacy Litigation, a consolidated case alleging Otter.ai recorded private conversations without consent from all participants and used them to train its models, according to court filings cited in HR Executive‘s April reporting. Both cases are still working through motions and class certification.
The Granola complaint focuses on the company’s own website copy, which the filing quotes as telling prospective customers that other people on a call won’t know it’s there and arguing that visible recording indicators change behavior in interviews and client conversations, according to the filing. Plaintiffs’ attorneys used that language to argue the lack of disclosure was a deliberate design choice, rather than an oversight.
Bradford Kelley, a shareholder at Littler Mendelson who has advised employers on the Otter.ai litigation, told HR Executive in April that the issue is drawing serious attention from clients. “The AI transcription and recording issue is a hot issue,” he said, adding that human resource teams should be “very interested in this case.”
Here is what both cases suggest for HR leaders managing AI tools inside interviews, performance conversations and internal meetings.
Consent law is not one rule
Recording consent rules vary by federal and state law, and employers should check every jurisdiction involved, according to Littler Mendelson’s analysis of the Otter.ai case. A meeting that includes employees or candidates in different states can trigger overlapping consent obligations that many employers have never mapped, Littler advises. The Granola complaint leans on California’s Invasion of Privacy Act specifically, which allows statutory damages of $5,000 per violation or three times actual damages, whichever is greater, PPC Land reported. The complaint describes a proposed class reaching into the millions, according to that reporting.
Opt-outs often protect the account holder
Model training is opt-out rather than opt-in on many notetaker plans, and the opt-out typically applies only going forward. According to the Granola complaint, the company’s own materials acknowledge it cannot confirm that a user’s data was excluded from training that happened before the setting was changed, and that isolating specific data from an already-trained model is not achievable with current techniques. The complaint also argues that the people whose voices get captured are frequently not the account holder at all. A candidate, or an employee on someone else’s call, has no setting to adjust.
Voice identification adds a biometric layer
Littler’s analysis of the Otter.ai case find that tools that attribute transcript lines to specific speakers by voiceprint may fall under state biometric privacy laws, including Illinois’ Biometric Information Privacy Act, which carries its own statutory damages. Granola’s complaint describes a similar speaker-attribution feature, according to PPC Land. Speaker-identification features may create biometric-privacy risk depending on how they work and what data they collect.
Banning the tools outright will not work
Littler says that employers cannot realistically prohibit AI notetakers and expect compliance. The firm points to survey data showing 1 in 5 professionals already use AI to draft meeting notes, meaning employees are bringing these tools to meetings whether or not a policy exists. Littler’s approach, per its analysis, is to select and configure a vetted tool, turn off voice identification where the biometric risk outweighs the benefit, put consent notices in front of every meeting rather than treating them as optional, set short data retention windows and write a clear policy on where these tools are and are not permitted, particularly in interviews, terminations and internal investigations.
Multinational employers face a stricter bar
Under GDPR, meeting recording and AI processing require a valid lawful basis and clear notice. Additionally, the EU AI Act may classify AI systems used for worker monitoring as high-risk, a category the firm says could reach notetakers offering sentiment or productivity scoring. In Germany, France and other co-determination countries, Littler advises that rolling out an AI notetaker can require works council consultation before deployment, a step with no real U.S. equivalent.
Neither the Otter.ai nor the Granola case has reached a verdict, and both companies dispute or have not yet publicly responded to the claims against them.
Credit: Source link
