AI Policy

Last updated: 11 October 2026

This policy explains how Meeti uses artificial intelligence, what data it processes, its limits, and your responsibilities. It forms part of our Terms of Service and Privacy Policy.

1. What our AI does

Meeti uses AI to convert meeting audio into text (transcription), detect the spoken language, and generate summaries, key points, decisions, action items, deadlines, and translations. AI output is a draft for you to review and edit.

2. AI providers and processing

  • Transcription and language detection are performed by OpenAI Whisper (model whisper-1).
  • Summarisation, analysis and translation are performed by OpenAI GPT‑4 (model gpt-4-turbo-preview).
  • Your audio and transcript text are sent to OpenAI only to produce these results for you.
  • Processing location: OpenAI's cloud infrastructure, primarily in the United States.
  • Under our OpenAI API agreement, content submitted through the API is not used by OpenAI to train its models and is not retained beyond the period OpenAI's API data-retention policy allows.

3. Your data is not used for AI training

We do not use your recordings, transcripts or notes to train third-party AI models, and we do not sell your content. AI processing happens solely to deliver the features you request.

4. Accuracy limits

  • AI output can be wrong: misheard words, wrong or missing action items, invented details, and incorrect deadlines.
  • Accuracy is lower for noisy audio, overlapping speakers, accents, and code-switching between languages.
  • Automatic language detection can misidentify the language.
  • Some languages advertised in the app are low-resource and may be unsupported or noticeably less accurate. Whisper reliably supports major world languages (for example English and Swahili); accuracy and support for Luganda, Runyankole, Rukiga, Ateso, Luo and Acholi vary and may be limited. Do not rely on output for these languages without checking the recording.
  • You must review AI output before relying on it, especially for legal, medical, financial, employment or safety decisions.
  • Meeti is not a substitute for official minutes or professional advice.

5. Human oversight

AI results are drafts you can edit. Our staff do not listen to your recordings unless you request support or a security/abuse investigation requires it.

6. Recording consent (your responsibility)

  • You must have the legal right to record every meeting you upload or capture.
  • You must inform participants and obtain any consent the law requires before recording.
  • Do not record private conversations without consent.

7. Prohibited AI uses

You may not use Meeti to:

  • record or process people unlawfully or without required consent;
  • generate content to harass, defraud or deceive;
  • process unlawful content, or content that exploits children;
  • attempt to extract, reverse-engineer or abuse our models or providers;
  • make solely automated decisions with legal or similarly significant effect on a person.

8. Third-party exports

When you export to Google Drive, Gmail, Google Calendar or Slack, you authorise Meeti to send that content to the service you choose. Those services' own terms and privacy policies apply.

9. Reporting AI problems

Report errors, harmful output or privacy concerns to [SET: support email]. We aim to respond within [SET: response timeframe, e.g. 5 business days].