Otter.ai
AI-powered meeting transcription and note-taking — fast and accurate, but your audio trains their models.
What should journalists know about Otter.ai?
Otter is the default transcription tool for a reason: it's accurate, fast, and the real-time meeting integration is genuinely useful for press conferences and routine interviews. But the privacy picture got significantly worse in 2025. A class action lawsuit (Brewer v. Otter.ai) alleges the company records conversations without proper consent from non-host participants and uses that data for AI training. A hospital breach in Ontario exposed patient health data through an unsanctioned Otter recording. The privacy policy still permits using your content to 'improve and develop' services. For routine, non-sensitive journalism work, Otter remains the most polished option. For anything involving confidential sources, use Good Tape (EU-hosted, no AI training) or local Whisper. The convenience isn't worth the exposure.
Press conferences, routine interviews, meeting notes, lecture transcription — anything where the content is not sensitive and speed matters.
Confidential source interviews, sensitive investigations, legally privileged conversations, any recording where content could put someone at risk if accessed by third parties. Also not ideal for non-English transcription (accuracy drops significantly).
Security & Privacy
Data is scrambled while being sent to their servers
Data is scrambled when stored on their servers
Where servers are located — affects which governments can request your data
Privacy policy summary
Otter stores audio and transcripts on US-based cloud servers. The privacy policy permits using your content to 'improve and develop' their services — which means AI model training. They claim data is de-identified before training use, but a 2025 class action lawsuit challenges whether de-identification is meaningful for voice data and conversational context. Otter's OtterPilot bot joins meetings automatically when calendar-integrated, and only seeks consent from the meeting host — other participants cannot opt out or disable recording. Deleted conversations go to Trash and are removed after 30 days, but there is no guarantee that data already used for model training is retroactively purged.
How to protect yourself:
If you use Otter for non-sensitive work: (1) delete recordings immediately after downloading transcripts, (2) never name confidential sources in any recorded conversation, (3) disable OtterPilot auto-join for meetings where you don't control the guest list, (4) use the desktop app rather than sharing transcript links, (5) review Otter's sharing settings — default behavior shares notes with all attendees. For sensitive interviews: use Good Tape (EU servers, no AI training, AES-256 encryption) or run OpenAI Whisper locally (free, fully offline, no data leaves your machine).
SOC 2 Type II and HIPAA compliance show genuine security investment, but the core problem is structural: Otter uploads all audio to US cloud servers and uses content for AI training. The 2025 class action lawsuit and 2024 hospital breach demonstrate real-world consequences of this architecture. Adequate for routine journalism. Not recommended for any work involving confidential sources or sensitive material.
Who Owns This
Known issues
Class action lawsuit (Brewer v. Otter.ai, filed August 2025) alleges Otter records conversations without consent from non-host participants and uses data for AI training — claims violations of ECPA, CFAA, and California privacy laws, ongoing as of April 2026. Ontario hospital breach (September 2024): OtterPilot auto-joined a medical meeting and transcribed PHI of seven patients, sent transcript to a former physician. Separate incident (September 2024): Otter captured sensitive post-meeting VC investor discussions and sent transcript to a participant who had already left. Privacy policy permits using content for AI model training — de-identification claims are contested since voice data resists true anonymization. OtterPilot joins meetings automatically via calendar integration; non-host participants cannot opt out, creating legal exposure in two-party consent states. No journalist-specific data handling commitments. No option to exclude data from AI training on Free or Pro plans.
Pricing
Free: 300 min/month, 30 min per conversation. Pro: $16.99/month ($8.33/month billed annually), 1,200 min/month. Business: $30/month ($20/month annually), 6,000 min/month. Enterprise: custom pricing (average ~$6,300/year per Vendr data). All paid plans include OtterPilot, AI summaries, and action item extraction.
None known.
This is an editorial assessment based on publicly available information as of 2026-04-02, using our published methodology. Independent security review is pending. Security posture can change at any time. This is not a guarantee of safety.
Something wrong or outdated? Report it.