AI Usage & Transparency

You interact with an AI system when you use Veritas Ask’s research and drafting functions. It supports research, comparison, extraction, document analysis, evidence summaries and preparation of draft work products. Legal, medical, academic and professional profiles guide the workflow; selecting a profile does not confer professional authorisation or regulatory approval.

The system combines your instructions, permitted conversation and workspace context, selected documents and, where enabled, external sources to generate text. It does not independently establish the legal truth of a matter or perform an official adjudication. A saved draft remains a draft until reviewed and adopted by the responsible professional.

Check that a cited source exists, that the quoted passage supports the statement, and that its date, jurisdiction, version and relevance fit the task. A reference, retrieval result or confidence description is not a guarantee of correctness. AI can fabricate a plausible citation, omit adverse authority, misunderstand a document or reproduce bias in its inputs. Assess counterarguments and missing information before relying on a conclusion.

Uploaded files may be parsed into internal sections. An internal excerpt used for context does not necessarily mean the original upload was incomplete. OCR, unreadable scans, unsupported layouts and context limits can affect extraction or coverage. Review reported extraction warnings and material source passages, particularly tables, handwritten text and exhibits. Ask for the scope of the review where full coverage matters.

For professional delivery, verify decisive facts and authorities, resolve uncertainty, make the necessary corrections and retain the sources and version relevant to your conclusion. The responsible professional decides whether the work is fit for its intended use. Do not present an unverified AI output as a verified finding or conceal legally required AI disclosure.

Confidential documents and personal data are not prohibited merely because they arise in professional work. You must be entitled to submit them, have a lawful purpose and legal basis, meet professional secrecy requirements, minimise inputs and use the applicable DPA and approved access controls. Use anonymised material where that achieves the purpose; pseudonymised information may still be personal data.

Special-category data, patient information, criminal-offence data and similarly restricted material require the additional statutory condition or authority and an agreed controlled deployment, with risk assessment and, where required, a DPIA. Document-level approval and an organisation’s internal sign-off help implement instructions but do not create legal authority or execute a supplier agreement. The same rules apply to sensitive content typed directly into a chat.

Classified information, passwords, private keys and information requiring an unprovided clearance or security regime must not be entered. The standard Service is not an approved classified-information system. Do not put sensitive case or patient details in public contact forms, feedback or public-source search queries.

We do not use Customer Content to train Veritas Ask. Inference is the processing required to produce your requested answer; retrieval makes permitted documents available to the task. Neither term means model training. Any supplier use must follow its agreed processing purpose and the applicable contractual restrictions; no customer opt-in to unrelated training is created by accepting the Terms.

Service telemetry records operational activity and usage. Abuse prevention and security logging protect the Service. Optional analytics measure permitted product interactions only with consent. Retention determines how long a record persists. These activities are distinct and remain subject to purpose limitation, the Privacy Policy and the DPA. The AI request setting that disables provider response storage is not a guarantee of zero retention across security logs and backups.

Customer Data storage and processing take place within the EU. Data Residency & Processing Locations and the applicable subprocessor schedule define the contractual boundary.

Legal workflows support research, drafting and professional analysis. They do not issue binding legal advice in the Provider’s name, judicial decisions or prosecutorial acts. Medical workflows support professional evidence research; the standard Service is not supplied for emergencies, autonomous diagnosis or treatment selection. Academic use must respect research integrity, authorship, confidentiality and applicable disclosure rules. A disclaimer cannot substitute for an assessment when the actual intended use changes.

Use by a court to assist research and interpretation of facts and law in a specific case, or certain law-enforcement uses, can require high-risk AI assessment. Human review alone does not remove that classification. Before such a deployment, the parties must establish the intended purpose, roles and applicable regulatory requirements in the institutional agreement. Standard access is not approval for a high-risk deployment or a claim of conformity assessment.

Do not use the Service for unlawful discrimination, prohibited manipulation, prohibited social scoring, unlawful biometric identification or other prohibited AI practices. Do not use it to make a solely automated decision producing legal or similarly significant effects without the required lawful basis, safeguards and a specifically approved deployment. This prohibition does not prevent lawful research into such topics.

The Customer appoints administrators and professional reviewers, approves data classes and sources, sets sharing and retention instructions and determines which outputs may be relied on. The Provider remains responsible for its own contractual and legal duties. These responsibilities cannot all be shifted to the end user through a disclaimer.

Report material output or source errors through the available feedback or contact channel using only the information needed to identify the problem. Report security or personal-data incidents to contact@veritasask.com and request a suitable secure exchange before sending confidential evidence. The Enterprise Terms, DPA and TOMs set the relevant assistance, incident and audit obligations.