Sellside Stack is reader-supported. Some links may be affiliate links: if you buy through them we may earn a commission, at no extra cost to you. It never affects our verdicts. How we make money.

No affiliate links on this page yet. If we add any, this page will be labelled Ad.

Disclosure: Sellside Stack may earn a commission if you buy through some links on this page, at no extra cost to you. Commissions never affect our rankings or verdicts, and vendors cannot pay to be included. Our research is independent. We have not hands-on tested the products in this article. This comparison is based on vendor documentation, trust centres, published pricing, security disclosures, court records and third-party reviews, all checked in November 2026 and listed under Sources. Nothing here is legal advice.

Otter vs Fireflies for confidential calls: which one can a deal team defend?

By Sellside Stack Editorial · Prices as of November 2026

Otter and Fireflies are the two names most people reach for when they want an AI bot to join a call and produce a transcript. For a sales stand-up, either is fine. For a management presentation, a lender call or a wall-crossing, the questions change: who consented, where does the audio go, who can read it later, is it used to train anyone's model, and is the vendor currently being sued over any of this?

On those questions the two tools are closer than their marketing suggests, but they differ in ways that matter. Fireflies has the stronger written governance story. Otter has the more serious open litigation and a training policy many compliance teams will not accept.

The short answer

Comparison table

Otter Fireflies
Price (as of Nov 2026) Free; Pro $8.33/user/mo annual ($16.99 monthly); Business $19.99/user/mo annual; Enterprise quote Free; Pro $10/seat/mo annual ($18 monthly); Business $19 annual ($29 monthly); Enterprise $39/seat/mo, annual only
Certifications SOC 2 Type 2 (security policies "based on" ISO 27001, not certified) SOC 2 Type II, ISO/IEC 42001
Training on your meetings Yes: Otter trains its own models on de-identified recordings and transcripts; its AI vendors do not train No: "never used for AI training"; zero-day retention agreements with AI vendors
Custom retention Enterprise Enterprise
Discard audio, keep text Enterprise ("audio removal from conversations") Enterprise ("transcript + summary only mode")
Consent tooling Enterprise: recording disclaimer email, MS Teams recording-permission capture Bot is visible; Chrome-extension capture without bot also offered
SSO / SCIM Enterprise, minimum 100-user licence Enterprise
Data location AWS S3 (US) US subprocessors (Google Cloud, MongoDB Atlas, AssemblyAI, OpenAI); private storage option on Enterprise
Open litigation In re Otter.AI Privacy Litigation (N.D. Cal.): wiretap, CIPA, BIPA claims proceeding; in discovery Fricker v. Fireflies.AI (N.D. Ill.): BIPA voiceprint claims; motion to dismiss pending

How we evaluate

For confidential deal calls we weight:

  1. Consent and visibility (25%). Can the firm prove every participant knew and agreed?
  2. Data minimisation (25%). Can audio be discarded, and can transcripts auto-delete on a schedule compliance sets?
  3. Training and secondary use (20%). Is meeting content used for anything beyond serving you?
  4. Access control (15%). Who sees transcripts by default, and can an admin override privacy settings?
  5. Legal exposure and cost (15%). Live litigation, and the real per-seat price of the plan you actually need.

We have not used either tool on live calls. Court-related statements describe allegations and procedural rulings, not findings of liability.

The rules that apply to both

Consent is your problem, not the vendor's. Otter's own privacy page says users "must always ask for consent and indicate when they are recording." In all-party-consent jurisdictions such as California, a visible bot alone is not proof that everyone agreed. Announce the recording, get spoken agreement, and keep it in the transcript.

MNPI changes the distribution question. A summary emailed automatically to every invitee, including the counterparty's advisers, or stored in a firm-wide searchable workspace, can carry inside information across a wall. Under UK and EU MAR, US Regulation FD and your firm's own information-barrier policies, who receives the transcript matters as much as who recorded it.

Litigation is about the vendor's own copy. The suits against both vendors ask whether the vendor kept and used meeting data for its own purposes, such as training or voiceprints. A vendor that discards audio and does not train has less to explain.

The reviews

1. Fireflies: the stronger governance story, if you buy Enterprise

Visit Fireflies →

Fireflies' Trust Center leads with "You own your data, and we never use it for training," backed by a "strict zero-day retention policy with AI vendors." It lists SOC 2 Type II and ISO/IEC 42001 certification (reports available on request), a GDPR DPA, and HIPAA and FERPA support for Enterprise customers who enable Private Storage. Its published subprocessors include AssemblyAI for transcription, OpenAI as the LLM, MongoDB Atlas and Google Cloud Platform, all US.

The controls a deal team needs are Enterprise-only: custom data retention (auto-delete), Transcript/Summary Only Mode, which keeps text and discards audio and video, Private Storage, SSO, SCIM, audit logs and a rules engine for routing meetings to channels. Enterprise is $39 per seat per month, annual only. Fireflies' help centre says AI credits (needed for AskFred and AI Skills) are not included in the Enterprise base price beyond a one-time 50-credit allowance, with add-ons starting at $5 per user per month. Pro ($10) and Business ($19) on annual billing do not include custom retention or transcript-only mode.

One control cuts the other way. Fireflies' Super Admin role "bypass[es] all meeting privacy restrictions within a workspace," including meetings marked "Only Me." That is useful for compliance supervision, but it means a deal team's private call notes are visible to whoever holds that role. Decide deliberately who that is.

Litigation: a first Illinois biometric privacy (BIPA) suit, Cruz v. Fireflies.AI, was voluntarily dismissed in March 2026 before Fireflies answered. A second, Fricker v. Fireflies.AI Corp. (N.D. Ill.), was consolidated with Martinez v. Fireflies.AI. A consolidated complaint was filed in June 2026, alleging that speaker recognition creates voiceprints without BIPA-compliant consent. Fireflies moved to dismiss in July and briefing closed in late August. The court has stayed discovery pending that ruling, and we found no decision as of early November 2026. A third-party tracker also reports a separate suit in the Northern District of California.

Workflow fit: A buy-side team that wants a searchable, text-only record of expert and management calls, auto-deleted after the period compliance sets.

Who it's for: Firms with an admin who will actually configure retention, sharing and recording rules. Skip it if: You are on Pro or Business and handling MNPI.

Cons: - Every control that matters is Enterprise-only, at $39/seat plus AI credits. - Super Admin can read "Only Me" meetings. - BIPA voiceprint suit still pending.

2. Otter: cheap and capable, but harder to defend

Visit Otter →

Otter is SOC 2 Type 2 audited, stores data on AWS S3 with AES-256 server-side encryption, and states that its third-party AI providers neither train on nor store customer data. Pricing on annual billing is $8.33/user/month for Pro and $19.99 for Business; Enterprise is quote-only. Otter's own pricing table puts the governance features on Enterprise: custom data retention, audio removal from conversations, a recording disclaimer email, Microsoft Teams recording-permission capture, recording control and sharing permissions, and export permissions. SSO and SCIM are listed with a minimum 100-user licence, which is a hard stop for a boutique.

The training question is the real problem. Otter's privacy FAQ says it "uses a proprietary method to de-identify user data before training our models." That is a clear statement that your meetings feed Otter's models, even if de-identified. Many investment-banking and legal compliance teams will treat that as disqualifying for client calls.

Litigation: in In re Otter.AI Privacy Litigation (N.D. Cal., No. 5:25-cv-06911), plaintiffs allege the Otter Notetaker joined meetings through calendar integrations and recorded participants without all-party consent, and that Otter used the content for training. On 13 August 2026 the court granted Otter's motion to dismiss only in part. Federal wiretap (ECPA), California Invasion of Privacy Act, Illinois BIPA, unfair competition and unjust enrichment claims proceed, while computer-fraud claims were dismissed. According to case trackers, Otter answered in September denying the allegations, and the class-certification motion is due in May 2027. None of this is a finding of liability, but the case will run well into 2027.

Workflow fit: Transcribing your own internal or public calls, such as earnings call replays, lectures or internal stand-ups.

Who it's for: Individuals and teams without client confidentiality obligations on the calls they record. Skip it if: The call involves clients, counterparties or inside information.

Cons: - Trains its own models on de-identified customer content. - Core privacy claims survived dismissal; discovery under way. - SSO/SCIM requires a 100-user minimum.

3. Granola: the no-bot alternative to both

Visit Granola →

If what worries you is a bot joining the call and the vendor keeping audio, consider a bot-free design. Granola captures device audio, deletes the audio after transcription on every plan, and is SOC 2 Type II certified. Business is $14/user/month. Enterprise ($35) adds workspace-wide auto-deletion, a training opt-out for the whole team, admin-enforceable notification that Granola is in use, and SSO (for workspaces of 50 or more users). By default it trains on anonymised data below Enterprise, with a per-user opt-out.

It is not a free pass. Chamberlain v. Granola, Inc. (N.D. Cal.), filed 30 July 2026, alleges that silent bot-free capture recorded a non-user without notice. Third-party trackers report no ruling on the merits yet. With no bot, nobody else on the call knows unless you tell them.

Configuration checklist for either tool

  1. Buy Enterprise or don't use it for MNPI calls.
  2. Turn off auto-join; start recording deliberately, per meeting.
  3. Announce and record consent at the top of every call.
  4. Set summary distribution to host only; never auto-email external attendees.
  5. Enable transcript-only or audio-removal mode.
  6. Set retention to your firm's policy, not the vendor default.
  7. Restrict Super Admin or equivalent roles to compliance.
  8. Keep deal calls out of firm-wide channels and workspaces.

FAQ

Is Fireflies safer than Otter for confidential calls? On published policy, yes. Fireflies states it never trains on customer data and offers transcript-only mode and custom retention on Enterprise. Otter trains its own models on de-identified data. Both need Enterprise configuration to be defensible.

Does Otter use my recordings to train its AI? Otter's privacy page says it de-identifies user data and then uses it to train its models. It also says its third-party AI providers do not train on customer data.

Can either tool delete audio but keep the transcript? Yes, on Enterprise only: Fireflies calls it Transcript/Summary Only Mode, and Otter lists "audio removal from conversations."

Are Otter and Fireflies being sued? Yes. Otter faces a consolidated privacy class action in California in which wiretap and biometric claims survived a motion to dismiss in August 2026. Fireflies faces Illinois BIPA claims with a motion to dismiss pending. Neither has been found liable.

What does a defensible setup cost? Fireflies Enterprise is $39/seat/month on an annual contract, plus AI-credit add-ons from $5/user/month. Otter Enterprise is quote-only, and SSO/SCIM needs at least 100 licences.

Verdict

For confidential deal calls, Fireflies Enterprise beats Otter. It has a clear no-training commitment, SOC 2 Type II plus ISO 42001, transcript-only mode and admin-set retention. The trade-offs are cost, AI credits sold separately and a Super Admin role you must lock down. Otter is good value for non-confidential transcription, but its training on de-identified customer content and the live wiretap and BIPA claims make it hard to defend for client or MNPI calls. If your real objection is the bot itself, look at a bot-free tool, and still say "this call is being transcribed" out loud.

Sources