How we protect your data and maintain privacy
Security and privacy are not features at Summit AI Notes—they are the foundation of everything we build. We've designed our architecture from the ground up to ensure your sensitive conversations and client data remain completely private and secure.
This document explains our security practices, architecture decisions, and what we do (and don't do) with your data.
All audio processing, transcription, and AI summarization happens on your Mac. Your meeting data leaves your device only if you switch something on yourself — a cloud AI provider, or iCloud sync.
We don't operate cloud servers for storing your data — there is no Summit backend at all. Recordings, transcripts, and notes live on your device. The one exception is iCloud sync: turn it on and your notes travel through your own iCloud account, encrypted with a key only you hold.
We have zero knowledge of your meeting content. We literally cannot access your recordings, transcripts, or summaries—they never touch our servers.
You have complete control over your data. You decide where files are stored, when to export them, and when to delete them. No hidden uploads or syncing.
All data is stored on your Mac using:
Audio transcription and AI summarization use local processing:
If you choose to configure cloud AI providers:
Summit AI Notes has minimal network requirements:
That is the whole list. The architecture review walks through each one with source citations; it was written for the Mac App Store version and does not yet cover the direct download's license-server calls.
It's important to be explicit about what we don't collect:
While Summit AI Notes is designed with security in mind, we recommend these additional practices:
Most tools rely on policies and legal agreements to protect meeting data. Summit keeps processing on your Mac instead: with Local Models selected, meeting audio and text are not sent to an AI provider. That can remove some third-party questions from a compliance review, but it does not make Summit compliant with any regulation on its own.
The EU AI Act regulates how AI systems process personal data. Summit's local-first architecture means your meeting data is never transmitted to external AI providers — no cross-border data transfers, no third-party processing, no data retention by AI vendors. Compliance is a byproduct of the architecture itself.
Local processing means personal data stays under your control. No data processor agreements needed for meeting content, no cross-border transfer mechanisms required, and no third-party sub-processors to audit.
Summit is a meeting notes app, not a medical scribe, and is not certified for HIPAA or any other healthcare regulation. With Local Models selected, transcription, summaries and chat run on your Mac. If you choose a cloud AI provider, it receives transcript excerpts and, for chat, your question with related topics, action items and meeting titles and dates, so your organization decides whether that provider is approved and what agreements it needs. The app does not separately encrypt its local database or recordings; they rely on macOS protections, so turn on FileVault. Check with your compliance team before recording meetings that include patient information, and see the healthcare page for what to confirm first.
Local processing helps maintain attorney-client privilege by keeping privileged communications off third-party servers. Financial services data residency requirements are met by default — your data resides where your Mac resides.
Note: Users in regulated industries should evaluate their specific compliance needs and consult with their legal/compliance teams before use.
Learn more about digital sovereignty for European professionals →We publish a full data-flow and security architecture review of the app: every trust boundary, every outbound connection the app can make, every privacy-relevant default, and the limitations we found reading our own code — each claim anchored to a specific file and line at a pinned commit.
Read the architecture review →We take security vulnerabilities seriously. If you discover a security issue, please report it responsibly:
We appreciate the security research community's efforts to keep our users safe.