AI Use Disclosure

1. What this Disclosure Is

This is a plain-language explanation of how NativeScan's AI features work and what happens to the content you submit. It is incorporated by reference into our Terms of Service and Privacy Policy and forms part of the agreement between you and us. By installing or using the App, you agree to this Disclosure and to the disclaimers and risk allocation in the Terms of Service. If you do not agree, you must not install or use the App.

2. The Provider Chain

• Your device prepares the payload. Text recognition (OCR) and the first pass of date detection happen locally, with no network connection.
• Our proxy, operated by us and hosted on Cloudflare, receives the payload over HTTPS. It is authenticated and accepts requests only from the App. It does not retain request or response bodies.
• Google runs the Gemini model that produces the output, under its API terms and Gemini data-use policies. We use the Gemini API rather than a consumer surface; this is the path on which Google's documented terms state that submitted content is not used to train its models.

We do not control these third parties and do not assume their obligations. A change in their terms or behaviour may affect what happens to content you submit. We may change AI providers at any time, on notice through an update to this Disclosure.

3. The Five AI Features

3.1 AutoName — suggests a document title

Transmits: one page image from the document, downscaled before sending, together with the recognised text of the document.
Runs: automatically after each scan finishes processing, and whenever you re-run it manually.
Limits: free tier has a weekly allowance (extendable by watching a rewarded ad); Pro is unlimited.

3.2 AutoSort — suggests a category folder

Transmits: nothing additional — AutoSort is part of the same single request as AutoName.
Runs: alongside AutoName. The suggested category is applied automatically for Pro subscribers.

3.3 Recall — ask questions about your documents

Transmits: your question, plus the title and an excerpt of the recognised text from up to five documents the App judges most relevant to it, plus recent messages from the current conversation so the model can follow context. If you dictate your question, see Section 6 — the audio may be processed by Apple.
Runs: when you send a message.
Limits: free tier has a weekly allowance (extendable by watching a rewarded ad); Pro is unlimited.

3.4 Key dates — finds due dates, expiry and renewal dates

Transmits: the recognised text of the document, plus today's date. This is the largest text payload of any feature — it needs to see much more of the document than naming does, because deadlines hide deep in the page.
Runs: automatically, in the background. The App first tries to find dates entirely on your device; only when that finds nothing does it send the text for AI extraction. On the free tier this applies to your two oldest documents; for Pro subscribers it applies to all documents.
Limits: deliberately does not consume your AutoName or Recall allowance.

3.5 Recall suggestions — proposes questions you might ask

Transmits: the title, category, and a short text excerpt from each of your three most recent documents.
Runs: automatically, once your library reaches three documents, and is refreshed periodically after that.
Limits: no allowance is consumed and there is no separate control for this feature.

4. Which Features Run Without Asking You

This is important enough to state on its own. AutoName, AutoSort, key-date extraction, and Recall suggestions all run automatically — after a scan, or in the background as your library grows. They do not prompt you each time. Only Recall itself is triggered by an explicit action every time.

There is no in-app switch to disable AI processing. NativeScan is designed and sold as an AI-assisted document scanner; the AI is not an add-on. Your acceptance of this Disclosure during onboarding is your consent to all of the processing described in Section 3. If you do not want a particular document's contents transmitted for AI processing, do not scan that document into NativeScan.

5. What is Not Transmitted

• Documents in your library are only transmitted when one of the features in Section 3 runs. Content is not continuously uploaded, and nothing is uploaded merely because it is stored in the App.
• Your real name, email address, Apple ID, or location is never attached to an AI request by us.
• The anonymous analytics identifier described in the Privacy Policy is not sent with AI requests. AI requests cannot be linked back to an install or to each other.
• Biometric data (Face ID / Touch ID / Optic ID) is handled entirely by iOS and never leaves your device.
• Neither we nor our proxy retain your documents, images, recognised text, questions, or the AI's answers after a request completes.

6. Voice Input

If you dictate a Recall question instead of typing it, the App uses Apple's speech-recognition framework. Depending on your device, language, and iOS settings, Apple may process that audio on Apple's servers rather than on your device. That processing is governed by Apple's privacy policy, not ours. We never receive the audio — only the resulting text, which is then transmitted as part of your Recall question. To keep audio entirely off the network, type your questions instead.

7. Important Limitations of AI Output

• AI can be wrong. Large language models are probabilistic. They can hallucinate, misclassify, mistranscribe, omit key details, and confidently state things that are not true. You must independently verify any AI output before relying on it for anything that matters.
• Extracted dates can be wrong or missing. A date the App shows you may be misread, misinterpreted, or attached to the wrong obligation — and a real deadline in your document may be missed entirely. Never use NativeScan as your only reminder for a payment, renewal, expiry, or appointment. Missing a deadline is your responsibility, not ours.
• AI is not advice. AI output is not legal, medical, tax, financial, immigration, accounting, psychological, regulatory, or professional advice of any kind. No advisor–client relationship is created by your use of the AI features.
• AI output is not confidential. It is produced by a third-party model. We do not represent that it is suitable for confidential, privileged, or regulated workflows.
• Output may reflect bias. Models can reflect biases in their training data. We do not warrant that AI output will be fair, balanced, complete, or appropriate for any particular use, group, or context.
• Availability is not guaranteed. AI features depend on third-party providers and may be slow, degraded, or unavailable at any time without notice.

8. Sensitive Content

NativeScan can scan many kinds of documents, including identity documents, passports, banking statements, financial records, medical records, immigration paperwork, legal documents, and employment records. The AI features rely on a third-party service. You are solely responsible for deciding whether any given document is appropriate to scan with an app that uses AI processing, keeping in mind that several features run automatically as soon as a document is scanned. Do not scan sensitive content with NativeScan if you are not comfortable with its content being processed by a third party. For workflows subject to legal or regulatory confidentiality requirements (for example certain HIPAA, GDPR, PHIPA, or solicitor–client privileged contexts), you should not use NativeScan without an independent legal review of your situation. We are not responsible for any breach of confidentiality, contract, professional duty, or law arising from your decision to scan a document into the App.

9. Do Not Rely on AI Output

You expressly acknowledge and agree that you will not rely on any AI output produced by the App for:

• any legal, regulatory, immigration, tax, financial, accounting, insurance, medical, mental-health, safety, security, or professional decision;
• any deadline, payment, renewal, expiry, or appointment — including any date the App extracts, displays, or sends you a reminder about;
• any filing, submission, application, or representation made to any government, tribunal, court, employer, professional body, lender, insurer, or counterparty;
• any decision affecting the rights, health, finances, residency, employment, safety, or welfare of yourself or any other person;
• any contractual, fiduciary, or regulated obligation; or
• any other use where an error in the output could cause meaningful harm.

You will independently verify any AI output before acting on it, and you accept that any consequence of failing to do so is your sole responsibility.

10. Risk Allocation

To the maximum extent permitted by law, you accept all risk for any AI output produced by the App and waive any claim against the Operator arising out of the content, accuracy, completeness, timeliness, suitability, lawfulness, or consequences of any AI output — including any claim based on negligence, and including any missed or misread date. Our total liability for any AI-related claim is subject to the limitation of liability set out in the Terms of Service.

11. AI Use is Mandatory

AI features are an integral part of NativeScan and are not separately optional. By installing and using the App, you agree to AI processing as described here, in our Terms of Service, and in our Privacy Policy. If you do not agree, you must not install or use the App, and you should delete it from your device.

12. Changes

If we change the AI provider, add an AI feature, change what data is transmitted, or change how it is transmitted in a way that materially affects you, we will update the version number and effective date at the top of this Disclosure and, where reasonably practicable, require renewed in-app acceptance.