AI Data Use Policy

Effective date: September 28, 2026
Last updated: September 28, 2026

This policy explains how Veskyra LLC handles information when you choose an AI-assisted content feature in its Services. It supplements the Privacy Policy, Terms of Service, and, for organizational processing where applicable, the Data Handling and Processing Agreement.

1. You choose when content AI is used

Content AI features are optional. An AI request begins only when you or an authorized user deliberately chooses an action identified as using AI. Opening an account, accepting our Terms, or saving information in a normal feature does not give us standing permission to submit that information to AI.

A selected action may process a document or a batch as described in the feature. For example, selecting an AI import for a PDF authorizes the extraction needed for that import; you do not have to approve each individual page separately unless the interface offers that control. We identify the scope before the task begins. A materially different task or expanded data scope requires an appropriate new instruction.

You may decline content AI and use available non-AI workflows. Some optional conveniences, such as automatic extraction from an image, may require manual entry if you decline AI. An organization may restrict which authorized users can access AI, but enabling the feature for an organization does not itself run it on all records.

This policy concerns AI used to process customer content. Authentication, fraud prevention, bot detection, and other security mechanisms operate as described in the Privacy Policy and are not optional content-assistance requests.

2. What the features do

Available features depend on the product and plan. These examples do not promise that every product supports every operation.

Kyrvec and Presvyr have no AI features; Presvyr's zero-knowledge design means we cannot read vault contents at all. Identity verification, barcode scanning, and other automated workflows in Kyrvec are not represented as AI features by this policy. If content AI is introduced in a future product change, it will require appropriate notice and remain subject to the explicit user-action and data-use protections in this policy.

3. Information submitted for a task

AI may process the selected text, images, files, instructions, relevant record context, and technical information needed to return a result. We limit task inputs to the requested function. The feature describes whether it uses a selection, a full document, associated record fields, or a batch.

Selected content may contain names, addresses, telephone numbers, injury or health details, confidential business information, regulatory records, metrics, or information about other people. It is not necessarily anonymized before processing. Remove unnecessary sensitive details where possible and submit only material you are authorized to process this way. Reskyr's intended users are non-transporting fire and rescue departments; Ryskra's are non-healthcare businesses. Those limitations do not make every incident record nonsensitive.

Choosing an AI action is an instruction to process the selected material, not proof that you have obtained every other person's required consent. Organizations remain responsible for an appropriate legal basis and any additional safeguards for sensitive or regulated data. HIPAA-regulated processing requiring a BAA is excluded from the standard service. A future exception requires a BAA and expressly approved service configuration; this policy does not represent current HIPAA readiness.

4. Hosted processing and no model training

The models used for these features run within the hosted infrastructure engaged to operate Veskyra's Services. The infrastructure provider processes requests to perform the selected task. We do not send those requests to a separate consumer chatbot account or authorize a model publisher to use customer content merely because its model is used.

We do not use customer inputs, uploaded material, or AI outputs to train, retrain, fine-tune, or improve AI models. We do not authorize providers to use that content for their own model training or independent service improvement. We do not sell it or use it for advertising.

Our infrastructure provider acts as our service provider for this processing. We may change the models or providers we use to improve quality, cost, or reliability, but any change will keep the commitments in this policy.

5. Temporary processing and saved records

AI needs to process task content to generate a result. A no-training commitment is different from a promise that no data is ever stored.

We do not enable provider content retention for independent model training or service improvement. Temporary task processing, queues, caches, and error handling must be limited to what is necessary for the requested feature and its security. Temporary inputs and outputs are kept only as reasonably necessary to complete the selected task, support a permitted retry, resolve a related processing failure, or investigate a specific security issue, and are then deleted or appropriately deidentified unless saved as an ordinary record through the selected workflow or law requires preservation. We determine the necessary period using task status, failure resolution, sensitivity, and any applicable legal or contractual requirement. We periodically review continued need; temporary processing is not a basis for an indefinite content archive. Specifically agreed shorter retention or deletion commitments continue to apply.

Source files you upload and results saved through the selected workflow remain ordinary account or organizational records. They follow the retention, export, and deletion rules in the Privacy Policy and applicable DPA. An AI import may intentionally save extracted fields as part of the action you selected. Stopping use of AI does not, by itself, delete those saved records.

We may maintain limited operational information such as time, task type, status, resource use, and security events to operate and troubleshoot the feature. We avoid including raw sensitive content where it is not needed. Authorized support or security personnel may access relevant information only for a permitted need and under applicable confidentiality restrictions. This is not routine human review of every AI submission.

6. Review results before relying on them

AI can introduce incorrect facts, omit information, misread an image, or change the meaning of a narrative. Check results against the source before using them. In particular:

• Verify names, dates, addresses, injuries, events, and other factual assertions in reports.
• Check certification names, license numbers, issuers, expiration dates, and renewal requirements against the original credential.
• Check extracted questions, answer choices, answer keys, formatting, and any accreditation claims before publishing training material.
• Do not allow an AI suggestion to replace a required professional, supervisory, or regulatory review.

AI output is assistance, not a guarantee of compliance, certification, accuracy, medical advice, or legal advice. These features are not intended to make final autonomous decisions about employment, eligibility, safety, or other similarly significant matters. Veskyra does not use these assistive features to make such decisions about you. You are responsible for reviewing results and for decisions made using them. AI results are provided under the warranty disclaimers and liability limits in Sections 8, 12, and 13 of the Terms of Service.

You and the relevant rights holders retain rights in submitted material. Rights in generated results are addressed in the Terms; AI output may not be unique or protectable by copyright.

7. Sharing and your controls

Using AI does not add an employer, training provider, or unrelated customer as a new recipient. A result saved into a record follows that record's existing permissions and applicable sharing rules. For example, a saved certification may be visible under a ValKred employer-paid plan or enabled sharing on an employer-sponsored discount plan, and imported test material may be visible to the provider's authorized team. Opting out of credential sharing is available for the discount arrangement, not the employer-paid plan; AI itself remains optional in both cases.

You may stop making AI requests at any time. Where a running task offers a cancellation control, you may use it; otherwise contact support. We cannot undo processing already completed. Request correction, export, or deletion of saved material through the relevant product tools or privacy@veskyra.com, subject to the record owner's rights and legal retention requirements.

If an employer controls a record, your ability to delete that organizational record may differ from your ability to stop using AI or change personal-vault sharing.

8. Changes and questions

We provide appropriate advance notice of material changes. We will not silently turn optional content AI into automatic processing of existing customer records or repurpose existing AI content for model training through a general policy update. Any new purpose must have a lawful basis and any required affirmative authorization before processing begins. Contractual restrictions continue to apply.

Questions about an AI feature or its data handling: privacy@veskyra.com. For assistance correcting an import or report, contact hello@veskyra.com.

Security incidents or suspected data exposure: security@veskyra.com.