AI Transparency Policy
Last Updated: September 20, 2026
Murmur uses artificial intelligence to help you capture and organize your thoughts. This policy explains, in plain terms, how our AI features work, what they do with your data, and what limitations you should be aware of.
1. How AI Is Used in Murmur
When you record a voice memory, your audio is sent to our backend, stored securely, and then processed using OpenAI's API to perform the following functions:
- Converting your spoken words into a text transcript
- Detecting the language you spoke in
- Translating your transcript when translation is needed or requested
- Generating a short, descriptive title for your memory
- Creating a semantic embedding of your memory, which is a numerical representation used to power search
2. What AI Processes
The AI processing pipeline works with:
- Your raw audio recording
- The transcript generated from that recording
This processing happens automatically as part of the normal memory-creation flow described in our Privacy Policy.
Your recordings and transcripts are sent for AI processing only after you have given your consent, as described in our Privacy Policy. You can withdraw that consent at any time.
3. What AI Generates
As a result of this processing, Murmur produces:
- A text transcript of your recording
- A translated version of the transcript, where applicable
- An AI-generated title summarizing your memory
- A search embedding used to help you find memories later
These AI-generated outputs are stored alongside your original recording so you can review, edit, or delete them.
4. AI Limitations
Artificial intelligence is a powerful tool, but it is not perfect. You should be aware of the following limitations:
- Transcripts may contain errors, particularly with unclear audio, background noise, overlapping speech, strong accents, or uncommon words and names.
- Language detection may be inaccurate for short recordings, mixed-language speech, or unusual phrasing.
- Translations may not fully capture tone, nuance, idioms, or context.
- AI-generated titles are automated summaries and may not always accurately reflect the full content or intent of your recording.
- Search results based on embeddings are approximate and may not always surface the most relevant memory.
We encourage you to review AI-generated transcripts and titles, especially for memories that are important to you.
5. No Human Review During Normal Operations
We do not manually listen to, read, or review your recordings or transcripts as part of normal operations. All processing described in this policy is performed automatically by software systems. Human review may occur only in limited circumstances, such as investigating a reported abuse violation, addressing a security incident, or complying with a legal obligation.
6. Training AI Models
We do not use your recordings, transcripts, or other personal content to train AI models unless we explicitly disclose this practice to you in advance and obtain your consent. If this changes, we will update this policy and notify you before any such use begins.
7. Your Control Over AI-Generated Content
You remain in control of your content at all times. You can:
- Review, edit, and correct transcripts and titles generated for your memories
- Delete individual memories, including their associated transcripts, titles, and embeddings (subject to the mandatory retention described in our Data Retention and Deletion Policy)
- Delete your account, which removes your AI-generated content in accordance with our Data Retention and Deletion Policy
- Withdraw your consent to AI processing at any time. Because transcripts, titles, and search all depend on it, withdrawing means we can no longer provide the Service to you (see our Privacy Policy)
8. Third-Party AI Provider
AI processing is performed using OpenAI's API. OpenAI acts as our Data Processor: it processes your data on our behalf, on our instructions, and solely to perform the functions described in this policy. OpenAI may process this data outside India, and we make such transfers in line with the DPDP Act and Rule 15 of the DPDP Rules. Our arrangements with OpenAI require reasonable security safeguards. Please see our Privacy Policy for more, including the circumstances in which the law may require us to disclose data to the Government.
9. Changes to This Policy
We may update this AI Transparency Policy as our AI features evolve. We will notify you of material changes through the app or other reasonable means.
10. Contact Information
If you have questions about how AI is used in Murmur, contact our Grievance Officer / Data Protection Contact. Details of the grievance process are in our Privacy Policy.
Grievance Officer / Data Protection Contact: [Insert name]
Email: subhrajyotisahoo08@gmail.com
Postal address: [Insert address]