Privacy
Without the asterisks.
Last updated · April 2026 · v2 'Tezcat' release
🔒 The core promise
LMIM OS is local-first by design. The model runs on your CPU or GPU. Your conversations, files, and configuration live in your home directory. There is no cloud account, no telemetry, no usage tracking — because there's no server on our end that would receive any of it.
This isn't a promise we have to enforce; it's how the software is built. The code is open source. You can audit every network call.
× What we never collect
- Your chat messages, prompts, or AI conversations
- Files you read, write, or operate on with LMIM
- Voice recordings, transcriptions, or generated audio
- WhatsApp, Telegram, email, Slack, or Discord message contents
- Calendar entries or meeting metadata
- IP addresses, geolocation, or device fingerprints
- Usage analytics, click tracking, or session recordings
💾 Where your data lives
Everything LMIM OS produces is written to your local filesystem. Nothing is synced anywhere unless you explicitly configure it.
You can delete any of these directories at any time. There is no remote copy.
☁️ Optional cloud providers
LMIM OS can route certain queries to third-party LLM providers (OpenAI, Anthropic, Groq) if — and only if — you explicitly configure an API key in ~/.lmim_os/.env and toggle Cloud mode in Settings.
When this happens:
- Your prompt is sent directly from your machine to that provider's API
- The provider's privacy policy applies to that prompt — not ours
- LMIM (the project) never sees the prompt or the response
- You can switch back to local mode at any time and the routing stops immediately
🌐 Website analytics
The lmim.tech website uses Plausible Analytics — a privacy-focused, cookieless alternative to Google Analytics — to count visits and understand which pages people read.
- No cookies and no tracking identifiers
- No personal data, IP addresses, or device fingerprints
- Only aggregated, anonymized counts
- EU-hosted, GDPR-compliant by design
Open your browser's network tab and verify — there are no third-party trackers.
💡 The "Shape v3" feedback form
If you submit feedback through the on-site suggestion form, here's exactly what happens:
- Your selected categories, free-text message, and optional email are sent to a single endpoint at
lmim.tech/api/feedback.php - That endpoint writes one JSON file per submission to a private directory on the LMIM server
- If you provided an email, it's used only to reply about your suggestion. It will never be added to a list, sold, or shared
- If you didn't provide an email, your submission is anonymous — there's no way to trace it back to you
- You can request deletion of any submission you made by emailing ops@lmim.tech
🏢 Enterprise deployments
For teams using LMIM OS Enterprise:
- All data remains within your infrastructure — we operate no backend
- Optional audit logs can be exported to your SIEM of choice
- Horus Guardian provides on-chain proof anchoring (Sepolia testnet) for compliance auditing
- We never access your deployment, your logs, or your users' data
For a detailed Data Processing Agreement, contact enterprise@lmim.tech.
📬 Questions or concerns
We're committed to transparency. If anything here is unclear, or if you want to verify a specific claim:
- Email: ops@lmim.tech
- GitHub: open an issue
- Mastodon: @lmim@fosstodon.org
LMIM OS is open source under the MIT License. The full source code is available at github.com/leanmeaninferencemachine — every claim on this page is verifiable in the code itself.