Safety Review Consent

This document explains the optional, revocable consent you may give NAL to allow a trained human safety reviewer to individually read your conversations — solely to detect and help prevent serious harm. It supplements, and must be read together with, the Privacy Policy. It mirrors the structure of the Data Improvement Consent.

1. What you are agreeing to

If you turn this consent on, you allow a trained NAL safety or clinical reviewer to individually read the content of your conversations only to detect, assess, and help prevent serious harm, including:

• self-harm or suicide risk;
• abuse, exploitation, or safeguarding concerns;
• credible threats of violence toward you or another person.

Nothing else. The reviewer reads for safety and for no other purpose.

2. This is separate from automated safety, which always runs

NAL's automated safety detection — for example the semantic crisis detection described in the Mira Addendum §3.3 — runs regardless of this consent, on every user, as a platform safety control. So does the emergency-disclosure path for imminent harm to life described in Privacy Policy §6 and Mira §4.1.

Turning this consent off does not turn off automated safety detection or the imminent-harm disclosure path. This consent only adds an optional layer of human review on top of them.

3. This consent does NOT cover Google user data

Human review under this consent applies to your conversations with NAL Services. It does not extend to any content sourced from your Google account (Gmail, Google Drive, or Google Calendar) — including summaries or quotations of it that a Service may produce. Content sourced from your Google account is only ever viewed by a human under the narrower "affirmative agreement to view specific messages, files, or events" condition set out in Google User Data & Limited Use §1 and §5. This carve-out is deliberate and is required by Google's Limited Use policy. (NAL does not currently offer any Gmail/Drive/Calendar feature — see that document §2 — so no such content exists in conversations today; this boundary binds us before any such feature launches.)

4. What this consent does NOT cover

• No AI model training. Content read under this consent is never used to train or improve AI models (consistent with Privacy Policy §4.1).
• No marketing, no profiling. It is never used for advertising, personalisation, or building a profile of you.
• No sale. It is never sold, licensed, or transferred to a third party for their commercial use.
• No Google user data, as stated in Section 3.

5. It is voluntary

Giving this consent is entirely voluntary. You are never required to enable it, and it is off by default. Whether it is on or off has no effect on your access to the service, your account, your wallet, or any feature you are otherwise entitled to use.

6. You can withdraw it at any time

You may turn this consent off at any time in your account settings (Profile → Privacy). Withdrawal takes effect going forward and does not affect your access to the service in any way. Once withdrawn, no new human safety review is performed under this consent.

7. How reviews are safeguarded

When a human safety review takes place:

• it is performed only by a trained safety or clinical reviewer bound by confidentiality;
• access is limited to what is necessary for the safety assessment;
• each review is recorded in an audit trail (who reviewed, when, why), so that access can be accounted for.

8. Imminent harm — independent of this consent

Where NAL believes in good faith that disclosure is necessary to prevent imminent harm to life, it may act under Privacy Policy §6 and Mira §4.1 whether or not you have enabled this consent. That emergency path exists to protect life and is not created, enlarged, or removed by this document.

9. Retention

Records created by a safety review (the reviewer's assessment and the audit-trail entry) are retained under the retention rules in Privacy Policy §8. Where a review relates to a clinical session with a linked verified therapist, the clinical-record retention rules in Mira §4.3 may apply.

10. Changes

If we materially change the scope of this consent, we will update the version number and the effective date above, and — where required — ask you to review the updated consent.

11. Contact