Skip to main content

AI Trust Passport

For AI vendors

Answer your customers’ AI due diligence once, with evidence a person has actually checked

If you sell an AI product to regulated or enterprise customers, you will know the pattern: a new security and AI governance questionnaire for every deal, the same evidence hunted down again, and answers that drift from one questionnaire to the next. The AI Trust Passport is a human-reviewed pack you can share with customers. It sets out what your AI product does, which controls we reviewed, the evidence behind each one, and where the gaps are. It is not a certification, and the front page says so.

Open

to AI vendors selling to regulated or enterprise customersCurrent status

10 areas

of AI governance, from intended use to decommissioning, scoped to your productWhat we review

Fixed fee

agreed in writing before we start, with the scope and exclusions set outNo day rates

Disclosed

where we helped you build or fix a control, the pack says so against that controlThe rule

Who this is for

Small and mid-sized companies that build or sell AI systems to organisations which run supplier due diligence: NHS bodies, public sector buyers, financial services firms, law firms and larger enterprises. It suits a vendor that keeps losing weeks to questionnaires, that has the evidence somewhere but not in one place, or that wants to know what a careful buyer would find before the buyer finds it.

If you are on the other side, assessing an AI supplier you are thinking of buying from, our vendor due diligence report is the right starting point instead.

What you receive

The AI Trust Passport

A dated document you can share with customers under your own terms.

  • The AI product and version, its intended use and the uses it is not meant for
  • The scope, the method and the framework editions used
  • Each reviewed control, its status and the evidence behind it
  • Open gaps, the reviewer’s name, the limitations and the date it should be reviewed again
The evidence register

Every document we relied on, so the next questionnaire takes hours, not weeks.

  • Owner, document date and version
  • Which controls each item supports
  • Its review status and classification
The gap list

What a careful buyer would question, in priority order.

  • Each gap, why it matters and who it falls to
  • A redacted version for sharing, with confidential attachments removed
  • Nothing is published or shared unless you choose to share it

What we review

Ten areas, scoped to the product in front of us rather than applied wholesale:

  • AI inventory, purpose, intended use and ownership
  • Governance, accountability and risk acceptance
  • Training and inference data, and privacy
  • Identity, access, secrets and infrastructure security
  • Model evaluation, robustness, bias and performance
  • Human oversight, explainability and transparency
  • Third-party models, dependencies and supply chain
  • Logging, monitoring, incident response and reporting
  • Change control, drift, retraining and lifecycle
  • Business continuity, resilience and decommissioning

Where it helps your customers, findings are mapped, with the edition stated, to the frameworks they ask about, such as ISO/IEC 42001, ISO/IEC 27001, the NIST AI Risk Management Framework and the OWASP guidance for large language model applications. Where the EU AI Act or UK law may apply, we say which obligations look relevant and that applicability is a legal question for a qualified adviser. We do not assume every AI product is high-risk.

How each control is reported

We do not give you a single “trust score”, because one number hides more than it shows. Each control carries one of these statuses, and the report never treats evidence received as evidence checked:

  • Not assessed
  • Evidence requested
  • Vendor asserted: you told us, but we have not seen evidence
  • Evidence provided: received, not yet reviewed
  • Human reviewed: adequate for scope
  • Human reviewed: gap
  • Not applicable, with the reason recorded

What it is not

The AI Trust Passport is not an ISO certificate, a conformity assessment, a regulatory approval, a penetration test, a legal opinion or a guarantee that your AI system is safe. It records what a qualified reviewer saw, on a stated date, within a stated scope. Your customers can read exactly that, and nothing more.

What actually happens, and when

First conversation

No charge. What you sell, who is asking you questions, and whether we have advised you before. If we have, we tell you now and introduce someone else.

Scoping letter

The fixed fee, the product and version in scope, the controls we will review, the exclusions, and a list of the evidence we will ask for. Evidence comes to us through a secure route agreed in the letter, never through the website form.

Your effort

Gathering the evidence, answering a structured questionnaire, and short conversations with the people who own the product, its data and its security.

Review and walkthrough

A named reviewer reads every item and records the reasons for each decision. We walk you through the Passport, the evidence register and the gaps. Fixing the gaps is your call; we will not sell you the fix.

Common questions

Is this a certification?

No. It is a dated, human-reviewed record of evidence within a stated scope. If a customer needs a recognised certification, such as ISO/IEC 42001, we will tell you and point you to an accredited certification body.

Will you help us close the gaps?

Yes, as separate work if you want it. The gap list is also written so your own team, or an adviser of your choosing, can act on it. Where we have helped close a gap, the pack says so.

Do you put our documents through AI tools?

No. Every review is done by a person, and your documents are not sent to external AI services. If that changes, it will be agreed with you in writing first.

Who can see the Passport?

You decide. Nothing is published. You receive a full version and a redacted version for sharing, and we keep a record of what was produced and when.

What does it cost?

The fee is fixed in writing before we start and depends on the product and the number of controls in scope.

There is no charge for the first conversation and no obligation at the end of it.

Standards we assess against

ISO/IEC 27001:2022 · ISO/IEC 27701:2025 · ISO 9001:2015 · ISO/IEC 42001:2023 · ISO/IEC 23894:2023 (guidance) · ISO/IEC 42005:2025 (guidance)

UK Cyber-Safe is an independent assessment practice. We are not a certification body and we do not issue certificates.

United Kingdom Cyber-Safe is a trading name of Eurologix Security Holding Group Ltd, registered in England and Wales, company number 11660724. Registered office: 128 City Road, London EC1V 2NX. Operating address: 1 St Peter’s Square, Manchester M2 3AE.