AI Transparency & Responsible AI Statement

TamRx uses AI to make regulatory work faster and more structured. We treat AI as an assistant to qualified judgement, not a replacement for it. This statement is written to inform rather than to market: it describes what the technology does, where it is limited, and what remains your responsibility.

• Classifying a regulatory question and routing it through our structured reasoning model.
• Drafting regulatory documents and memos from a curated, reviewed knowledge structure.
• Simulating the perspectives a reviewer might take, to help you anticipate questions.
• Summarising and organising regulatory considerations into a structured assessment.

To do this, we use large language models provided by OpenAI and Anthropic, combined with our own structured reasoning layer (the Regulatory Transition Network) that grounds and constrains the output.

• We do not provide legal, medical or regulatory advice, and the AI is not a qualified professional.
• We do not make or guarantee regulatory decisions, classifications or approvals on your behalf.
• We do not use your assessment inputs to train TamRx’s own foundation models.
• We do not present AI output as a substitute for review by you, your qualified personnel, a regulatory consultant, a notified body or a competent authority.

• Hallucination. Language models can generate plausible but inaccurate statements, citations or requirements. Verify any specific obligation, date, standard or citation against primary sources.
• Currency. Regulations and guidance change, and some areas — such as the EU AI Act timeline — are actively evolving. Output may not reflect the most recent position; confirm current requirements before relying on them.
• Confidence is not certainty. Confidence levels, probabilities and scenario rankings are indicative aids to prioritisation, not predictions or guarantees.
• Context limits. The Service reasons only from the information you provide and its curated knowledge structure. Incomplete or imprecise inputs produce less reliable output.
• Not a substitute for judgement. Regulatory determinations depend on facts and context the Service cannot fully assess.

Every output is intended to be reviewed by a competent person before use. You are responsible for verifying output against primary legislation, official guidance and qualified professional judgement, for confirming the accuracy of any classification or obligation, and for approving any document before it is submitted, relied upon or shared. The Service supports this by citing primary sources and by structuring output so it can be checked, but the review itself remains yours.

We ground AI output in a curated, reviewed reasoning structure rather than relying on free-form generation alone; we cite primary sources so claims can be checked; we present indicative confidence rather than implying certainty; and we design the Service to operate on de-identified, product- and device-level information. These measures reduce, but do not eliminate, the limitations described above.

We are mindful that AI used in regulated settings is increasingly subject to its own requirements, including under the EU AI Act. TamRx is a decision-support tool for regulatory professionals; it is not a medical device and is not intended to provide a clinical function. Where you use TamRx in connection with your own regulated product, responsibility for that product’s compliance remains with you.

Provide accurate inputs; avoid submitting identifiable patient data; review and verify every output; confirm current requirements from primary sources; and ensure a qualified person approves any document before use. See our Terms of Service for the full allocation of responsibilities.

Questions about how TamRx uses AI can be sent to legal@tamrx.co.uk.