Fathom Recording: Permissions, Storage and Supported Meetings

Fathom recording supports Zoom, Google Meet and Microsoft Teams, but the exact capture method now matters. Fathom’s newer experience documents bot-free transcript-only and audio-only modes as well as an audio-and-video mode that can use a meeting bot. The previous experience is primarily bot-based. Availability differs by device and rollout.

This article uses Fathom’s official help centre checked on 19 July 2026. The product was changing rapidly in 2026, so confirm the current experience, operating system and account settings rather than relying on an older walkthrough.

What Fathom recording is designed for

Fathom turns supported online meetings into a transcript, recording or summary that can be reviewed and shared. Its compatibility page lists standard Zoom, Google Meet and Microsoft Teams links on Mac and Windows. It does not currently describe general cellular-call recording as a supported use case.

That makes the intent distinct from a universal voice recorder. Fathom is strongest when a conversation has a supported meeting link and the user wants call-specific AI follow-through. The Fathom transcription guide focuses on the text output; this page focuses on capture, permissions and storage workflow.

Current capture modes

Fathom’s new-experience documentation describes four choices: Capture Off, Transcript Only without a saved recording, Audio Only captured through a bot-free mode, and Audio & Video using a bot. Fathom has also documented beta work on botless video recording, so the interface may continue to change.

Do not describe Fathom as always bot-based or always bot-free. Open the pre-meeting panel, inspect the selected mode and verify what will be retained. A transcript-only mode can reduce stored media but still processes conversation content; “no recording saved” does not mean “no data processing.”

Supported devices and meeting limits

Fathom currently documents Mac and Windows support. Its compatibility article excludes Chromebooks, Linux, mobile devices, tablets, phone calls without a standard meeting link, webinars and breakout rooms. The newer bot-free experience was documented as Mac-only in May 2026, with Windows support planned.

These details are operationally important. A team cannot infer that a colleague on Chromebook has the same workflow as a Mac user. Test each supported platform and keep an alternative for unsupported devices. The Fathom reviews article provides a broader product-fit assessment based on official evidence.

Participant notice and consent

Fathom says there is no silent recording path: the previous experience uses a visible bot and platform notices, while the newer experience can use a meeting-chat Recording Notice or in-app consent prompts. Its consent guidance also describes optional emails to external attendees for qualifying scheduled meetings.

Notice is not the same as valid consent in every context. Fathom explicitly places responsibility on the recording owner. Explain the purpose, capture mode, recipients and retention, ask for agreement and offer manual notes if anyone declines. Same-day or impromptu meetings may not receive advance consent emails, so never depend on automation alone.

Fathom is optimized for supported online calls; Kuno is built for authorized conversations in the room. Kuno is a physical AI voice recorder designed and developed in Munich, with EU-hosted processing and storage. Explore Kuno

Host permissions and failed capture

Meeting-platform permissions still matter. Waiting rooms, host controls, administrator policy and meeting-link format can prevent a bot or capture flow from starting. Zoom consent-email behavior can also depend on host privileges. A Calendar event alone is not proof that Fathom will record successfully.

Before an important call, confirm the Fathom panel sees the meeting, the intended capture mode is selected and the recorder is active only after consent. If a bot requires admission, assign someone to admit it. Keep manual notes available. Do not restart capture secretly after a participant or host has declined it.

Recording ownership, sharing and access

Fathom can auto-share a summary and recording, a summary only, or nothing, depending on settings. Its auto-sharing documentation warns that Calendar invitees may receive access even if they did not attend, and that an already sent email cannot be unsent, although access can be revoked.

Review defaults before sensitive calls. Avoid “anyone with the link” when named access is appropriate. Confirm whether internal calls become visible to a team and who owns the recording after an employee changes role. A convenient automatic email can become a disclosure incident if the invitation list was broader than attendance.

Retention, download and deletion controls

Fathom’s organization settings include retention periods and options to disable recording download or deletion for team accounts. Those controls can support governance, but they must reflect an approved policy. Preventing deletion indefinitely is not automatically safer than a proportionate retention schedule.

Document who can set retention, who can export and how participant requests are handled. Test deletion and access revocation before deployment. If the organization requires a particular processing region or contractual protection, verify it directly in current legal and security documentation rather than inferring it from a feature page.

Transcript and summary accuracy

Fathom’s transcript and AI summary remain machine-generated outputs. Verify names, numbers, product terms, decisions, owners, deadlines and negation against the authorized source. Test real language combinations and audio conditions. Fathom’s language count or feature availability does not guarantee equal performance for every accent or domain.

Do not automatically sync unreviewed action items to a CRM. Approve the final record first, especially for sales commitments, employment discussions, legal matters or financial forecasts. A defined meeting-minutes format helps distinguish generated notes from an accountable record.

In-person meetings are a different use case

Fathom’s official in-person help article described a workaround for the previous experience: create a supported online meeting on a computer and position the laptop to capture the room. It warns that speakers in the same room may all be attributed to the Fathom user. Fathom also announced planned iOS support, but planned functionality should not be treated as generally available.

For regular room meetings, compare a physical recorder, phone and laptop under real conditions. Evaluate microphone coverage, visible control, speaker handling and the data path. The voice recorder for interviews guide covers consent and source-quality considerations for face-to-face work.

How to test Fathom recording

Use a normal user account for a consented pilot across Zoom, Meet and Teams if all three matter. Include external guests, waiting rooms, a same-day meeting and a meeting where recording is declined. Confirm notice, capture mode, transcript creation, sharing recipients, export, access revocation and deletion.

Measure missed captures, correction time and participant friction. Verify that the free or paid tier used in the test matches deployment. Product updates can change mode names or availability, so keep a dated internal runbook and review it after major Fathom releases.

Best-fit conclusion

Fathom recording is a good fit for people whose work happens on supported online platforms and who want transcription and summaries integrated with capture. Its evolving bot-free modes can reduce reliance on a visible participant bot, but they do not remove consent, account or data-governance responsibilities.

It is a weaker fit for unsupported devices, normal phone calls, webinars, breakout rooms and recurring in-person meetings. Choose the exact mode intentionally, verify participant notice and keep every AI output subject to human review.

Cover consented meetings that do not have a Zoom, Meet or Teams link. Kuno supports physical room capture and reviewable AI drafts for in-person work. See Kuno