Trust and legal / Public summary
AI Use Policy
AI features assist staff with questions, adverse-event narratives and sponsor email drafts, and provide an automated participant prescreen chat.
A named person reviews and approves AI drafts before use; AI never decides eligibility, signs or sends anything on its own.
Every AI request and output is logged with the model version and approving person, and customer data is never used to train AI models.
AI features use a third-party model provider; the current provider is listed in our subprocessor notice on request.
The prescreen is an automated assistant, not medical advice, and only study staff decide participation; you can reach a human through your research site's contact details or privacy@kyrostrials.com.
For an emergency, call 911 rather than using the assistant.
State AI notices
California
For communications covered by California AB 3030, AI-generated clinical communications require an AI disclosure and instructions to contact a human provider unless the applicable licensed-provider review exception applies. Contact your research site for a human provider, or privacy@kyrostrials.com for help reaching the site.
Texas
People receiving health services should be told when AI is used by the time the service is first provided, or as soon as reasonably possible in an emergency, where applicable. Notice: This site uses AI-assisted tools, reviewed by study staff. Contact your research site for a person.
Utah
When asked, an AI interaction must be disclosed where required. Regulated occupations may also need prominent disclosure at the start of high-risk interactions involving sensitive information. The prescreen identifies itself as an automated assistant at the start; contact your research site to speak with a person.
The full agreement is in final legal review. Request a copy at legal@kyrostrials.com.
Request a copyReader guidance, separate from the document
This section explains how to review AI review. It does not amend the policy or supply missing contract terms. Read the published document above and request the current full agreement where the page is a summary. Your organization should confirm that the wording addresses the intended use before relying on it.
Begin with a list of disclosure, configured availability, draft review and output handling. Identify who at the site owns each question and what evidence is needed. A general product description cannot answer every contractual or study-specific issue. Keep unanswered questions in the review record rather than treating silence as a favorable term.
Review the workflow and the evidence
The practical responsibility is a named person’s decision before using an AI draft. Ask what record would show that the process was followed and who can review it. A demonstration with sample data can clarify the workflow without disclosing participant information. It does not replace your legal or operational assessment.
For electronic research records, inspect the responsible user, dates, status, record linkage and change history. For public website interactions, review the information requested and the stated purpose. These are different contexts; a business-contact form should not be used to submit source data or confidential study documents.
Keep access and communication specific
Use individual accounts and the access appropriate to the person’s role. A shared login prevents a reliable link between an action and its responsible user. When inviting a reviewer, check the study and access period rather than assuming that a broad organization role is necessary.
For a question about this document, contact legal@kyrostrials.com. For privacy rights or handling concerns, contact privacy@kyrostrials.com. For a technical barrier or service problem, contact support@kyrostrials.com. Include business contact details, the page and the question, but do not paste participant names, health information, credentials or signed study documents into ordinary email.
Plan review before sensitive use
Request the current security summary, provider information and relevant agreement status before transferring sensitive records. Public summaries do not establish BAA execution, a particular vendor arrangement, an uptime guarantee or a certification. If a requirement is unresolved, document it and obtain an answer instead of inferring readiness from a feature being visible.
Your site’s procedures should explain who approves the system’s intended use, how staff are trained and how exceptions are handled. Keep the review proportionate to the study and data involved. A software control supports the procedure; it does not replace investigator responsibility or participant-facing obligations.
Retain context when records move
Before importing or exporting study information, identify the original records, signature evidence and audit histories that must remain available. A spreadsheet can carry values and dates without reproducing the source system’s historical events. Keep source exports and reconcile sample records so a later reviewer can understand the transition.
Review the trust hub for related policies and summaries and the switching guide for a sample migration review plan. Contract-specific archive or exit terms need the full agreement. Customer data is not deleted for non-payment, but your organization still needs to plan record retention and access under the applicable research rules.