ASSUR.AI is a membership and assurance framework for businesses that want to demonstrate that their use of AI is responsibly governed. It combines public commitment, a published Standard, evidence-based assessment and a verifiable Trust Register.
ASSUR.AI is a membership and assurance framework for businesses that want to demonstrate that their use of AI is responsibly governed.
Members make a public commitment through the ASSUR.AI Charter and can progress to an evidence-based assessment against the published ASSUR.AI Standard. Businesses that meet the assessment requirements earn the ASSUR.AI Trustmark.
Every member appears on the Trust Register. The point is not simply to make a claim about responsible AI, but to make the business's current standing independently checkable.
Our mission is simple: responsible AI, made provable.
Clients, insurers, procurement teams and regulators increasingly want clearer answers about how businesses use AI. Saying that AI is used responsibly is one thing. Showing that responsibility is governed, documented and checked is another.
ISO/IEC 42001 provides a comprehensive route for businesses seeking a full AI management system. ASSUR.AI addresses a different need: a proportionate assurance route designed primarily for smaller and medium-sized businesses that need credible governance without reproducing a full enterprise compliance programme.
Proportionate does not mean minimal. The requirements still have to be meaningful to the clients and other stakeholders who rely on the result.
The ASSUR.AI Charter and Standard were developed with AI researchers and experts at the Oxford Centre for Artificial Intelligence.
The requirements are informed by established approaches including ISO/IEC 42001, the NIST AI Risk Management Framework, UK regulatory guidance and the EU AI Act. "Informed by" is deliberate: ASSUR.AI does not claim that its Standard substitutes for any of them.
The Standard is organised around eight areas of responsible AI governance and uses fifteen published assessment criteria as the current baseline. It is version-controlled so that changes can be communicated and applied consistently.
The Oxford Centre for Artificial Intelligence contributes research and technical expertise to the development and continuing review of the Standard.
Its role is to help keep the requirements connected to developments in AI, regulation and professional practice. It does not determine the outcome of individual member assessments.
ASSUR.AI remains responsible for membership, assessment decisions, the Trustmark and the Trust Register.
Paying for membership or assessment does not buy the outcome. Every business is assessed against the published requirements and every final assessment decision is made by a person.
Assessment decisions are based on the evidence submitted. A firm's size, commercial relationship with ASSUR.AI or route into the network does not change the assessment criteria. ASSUR.AI records the basis of assessment decisions and applies procedures for conflicts of interest and review where they are needed.
Where a decision affects a member's standing, the member is told the reasons and can use the review route set out in the Membership Terms.
ASSUR.AI provides assurance against the ASSUR.AI Standard. It is not accredited certification and does not claim to be.
The Trustmark makes a deliberately specific claim: the business has provided evidence against the published ASSUR.AI requirements, that evidence has been reviewed through the assessment process, and the business holds the status shown on the Trust Register.
It does not claim that every use of AI is risk-free, approve a particular AI model or supplier, or establish compliance with every law or regulatory obligation that may apply.
ASSUR.AI, ISO/IEC 42001 and the EU AI Act are different kinds of instrument. ISO/IEC 42001 is a voluntary AI management-system standard. The EU AI Act is law where it applies. ASSUR.AI is a voluntary assurance framework with a published Standard, an assessment process and a public Trust Register.
The requirements overlap in important themes such as accountability, inventory, risk, transparency, data governance, human oversight and review. Work done for one may therefore be useful elsewhere, but ASSUR.AI does not claim formal equivalence, credit or exemption.
Nothing voluntary discharges a legal obligation. Businesses remain responsible for understanding the laws, professional rules and contractual duties that apply to their particular use of AI.
ASSUR.AI is designed primarily for smaller and medium-sized businesses in professional services and adjacent sectors where clients expect judgement, confidentiality and accountable decision-making.
It is aimed at businesses that use AI in their work rather than businesses whose principal product is the development and sale of AI systems. Larger businesses may have needs better served by a full management-system approach or other sector-specific assurance.
Membership belongs to businesses rather than individuals. “Who is covered”, in the Membership chapter, explains how group structures and trading entities are treated.
Membership is the starting point of the ASSUR.AI journey. Businesses join by signing the Charter, gain access to the Standard, training and member resources, and can progress towards independently assessed assurance.
The ASSUR.AI Charter contains ten public commitments on how a business uses and governs AI. A named accountable person signs it on behalf of the business.
The Charter is public. It sits behind the business's Trust Register entry so that clients can see what has actually been committed to rather than relying on a generic responsible-AI statement.
Signing is the foundation of membership. The formal register distinguishes Member and Signatory status; the membership journey may describe the early stage more simply as becoming Committed.
Training helps turn the Charter from a signature into something people across the business can understand and apply.
Required introductory training forms part of the membership pathway before a business progresses to assessment. ASSUR.AI identifies the required participants during onboarding and records completion against the member account.
Training does not itself create an assurance result. The Trustmark is earned through evidence-based assessment.
Membership is charged annually, based on firm size, and covers 12 months from your sign-up date. The fee is payable when you join and includes assessment and annual confirmation to maintain your membership standing. Renewal falls on the anniversary of your sign-up date, rather than the calendar year.
All fees exclude VAT.
Assured, the future level, will be priced separately. Its price will be published before it becomes available.
From the day you join, membership includes practical tools, training and resources to help you govern AI responsibly and prepare for assessment.
The badge is not a Trustmark. A Trustmark is earned only after the relevant evidence has been assessed.
Assessment adds more than the Trustmark itself. Assessed members also benefit from:
Membership and assurance apply to the legal entity identified on the Trust Register. That entity is the business entitled to use the badge.
Branches operating as part of the same legal entity and under the same governance are covered by that membership. Separately incorporated subsidiaries, sister companies, franchises, joint ventures and other legal entities are not covered unless they hold their own membership and Trust Register entry.
A trading name may refer to the member's status only where it is clear which legal entity holds the membership. Group-wide claims must not be made where only one entity holds the relevant standing.
Membership runs for 12 months from the date it begins and is renewed on the anniversary of that date. ASSUR.AI sends a renewal reminder before the end of the paid period and members confirm that the information and commitments behind their standing remain current.
If membership or assurance standing lapses, the Trust Register records the current position rather than leaving an expired claim looking valid. Rejoining is possible, but historic standing is not backdated or presented as continuous.
Payment, expiry, suspension and reinstatement are governed by the Membership Terms.
The Charter runs both ways. Members make public commitments, and ASSUR.AI maintains the Trust Register as the public record of current standing, applies the published assessment requirements consistently and keeps the distinction between membership badges and earned Trustmarks clear.
ASSUR.AI publishes the material requirements used for assessment, gives reasonable notice of material changes, treats assessment evidence as confidential subject to the Membership Terms and Privacy Policy, and provides a route to question decisions affecting standing.
The outcome of assessment is public. The confidential evidence submitted to support it is not.
ASSUR.AI is opening with a selected first cohort of founding members. Their role is to establish the network at launch and provide practical feedback while the member experience is still being refined.
Founding status does not lower the assessment bar and does not buy an assurance outcome. Any founding-member privileges are kept separate from assessment decisions and are published alongside the conditions that apply to them.
Assessment is how businesses move from commitment to demonstrated assurance. Members provide evidence against the published ASSUR.AI Standard, which is reviewed against defined criteria before Assessed status and the Trustmark can be awarded.
ASSUR.AI is a progressive journey towards responsible AI assurance, rather than a collection of separate credentials.
The three formal stages of membership are Charter Signatory, Assessed and Assured.
Charter Signatory recognises participation and a public commitment to responsible AI use. Assessed is awarded only when members demonstrate compliance with the published Standard through a formal evidence-based assessment. Assured represents a planned higher level of assurance, with detailed requirements to be published ahead of launch.
Each stage builds on the last. Trustmarks must be earned through progression and assessment; they cannot be purchased or stages skipped.
The fifteen assessment criteria turn the Charter commitments into items a business can evidence. They cover accountability, AI inventory and risk, policy and decisions, transparency, data and inputs, operational controls, incident response and periodic review.
The full criteria are available to members.
The criteria are a baseline for responsible governance. They do not remove a business's responsibility to meet laws, professional rules or contractual obligations that apply to its specific work.
The inventory covers machine-learning systems used by the business, including AI features embedded in third-party products. Low materiality is handled through the risk level rather than by leaving the system off the inventory.
Examples include generative assistants, transcription and meeting-summary tools, document review and semantic search, screening or scoring systems, chatbots, forecasting tools and machine-learning features within everyday business software.
Ordinary deterministic software, such as a spreadsheet formula or rule-based workflow, is not treated as AI merely because it is complex. Where a vendor cannot confirm whether machine learning is involved, the member records the enquiry and the response as part of its evidence trail.
When a member is ready, it provides evidence against the fifteen published assessment criteria. Evidence may include policies, an AI inventory, risk records, decision records, governance arrangements, incident procedures and other artefacts that show how the requirement is met.
Automated tools may support administration by checking submissions for completeness, organising evidence and flagging items that may require closer review, but they do not make assessment decisions. Every final assessment decision is made by a person.
If a submission does not yet meet the requirements, the assessment identifies the gaps that need to be addressed. Businesses can then submit additional evidence. A successful assessment results in Assessed status and the Trustmark being recorded on the Trust Register.
The current assessment establishes that the required governance has been evidenced. Assured, the future level, will make a higher assurance claim once its requirements and assessment process are published.
Assessment may require commercially or operationally sensitive material. ASSUR.AI does not publish assessment evidence on the Trust Register. The public record shows the outcome and the information needed to verify the member's current standing.
ASSUR.AI treats submitted evidence as confidential and restricts access to people and service providers who need it for assessment, member support, security or legal obligations. Identifiable assessment evidence is not used for research or marketing without permission.
How evidence and related personal data are handled, how long records are kept and when information may need to be disclosed will be set out in the Membership Terms and a member privacy notice before assessment opens.
Assurance standing is time-limited because AI use, internal controls and the Standard itself can change. Members therefore complete an annual confirmation as part of renewal.
The annual confirmation asks about material changes since the previous assessment or confirmation. Where a change affects the basis on which a status was awarded, ASSUR.AI may request further evidence or require reassessment.
The annual confirmation for the current pathway is included in the published annual membership fee.
Decisions affecting a member's standing come with reasons and a route to request review, except where immediate action is necessary to protect the integrity of ASSUR.AI.
A member may request a review in accordance with the Membership Terms. Wherever practicable, the review is handled by a person who was not the original decision-maker, and the member receives the outcome and reasons in writing.
At launch, this is an internal review process. ASSUR.AI may introduce an independent external review element as ASSUR.AI develops.
Verification is what makes ASSUR.AI publicly checkable. Every ASSUR.AI badge is linked to a live Trust Register entry showing the business, its current status and the assurance it has earned.
The ASSUR.AI badge and the ASSUR.AI Trustmark make different claims.
The badge is an informal marker of membership. It shows that the business is part of ASSUR.AI and has made the public Charter commitment. It is not evidence that the business has passed assessment.
The Trustmark is earned through assessment. The Assessed Trustmark means the business has provided evidence against the published ASSUR.AI requirements and met the relevant assessment threshold. Assured, the future level, will carry a higher claim when its requirements are launched.
A mark is useful only if someone can check the claim behind it. Every ASSUR.AI badge displayed online must link directly to the member's live Trust Register entry.
Where a live link is not possible, such as on printed material, the membership number must be displayed clearly with the mark. The Trust Register is the definitive record of the member's current standing.
Static or copied marks can become out of date. The register shows the status that applies now.
Members may use only the badge that corresponds to their current standing and only for the business to which it was issued.
Marks must remain legible and must not be altered, combined with other graphics or presented in a way that changes or exaggerates their meaning.
Digital and print use is governed by the Badge and Trustmark Use Rules, which form part of the Membership Terms.
A Trust Register entry identifies the member business, its current formal status, membership number, relevant dates and the public Charter or accountable-person information associated with that membership.
Each business is assigned one membership number when it joins. That number remains with the same legal entity through renewal, progression, lapse and rejoining. It is not transferred to another business and does not change when the member moves between statuses. A new legal entity requires a new membership and number.
A membership number looks like GB-2026-7X4K-M6. The status shown is one of the three formal stages of membership: Charter Signatory, Assessed or Assured.
The register is a neutral record of current standing with ASSUR.AI. It publishes the statuses and achievements currently held, not the confidential evidence supplied during assessment or unsuccessful assessment attempts. Where a status is no longer current, it is not presented as a current achievement, and the relevant dates make clear the period for which standing applied.
For group structures, the relevant entry is the entry for the legal entity whose status is being relied on, so users can verify the correct entity directly.
Anyone can verify a claim without setting up an account. Search the public Trust Register by business name or membership number, or follow the link carried by a digital badge.
All routes lead to the live register entry. If a badge appears to disagree with the Trust Register, rely on the Trust Register and contact ASSUR.AI if clarification is needed.
A badge is not a Trustmark, and a Member or Signatory entry must not be presented as though it represented an Assessed result.
A credible register needs a route for concerns about inaccurate entries, misuse of a badge, or conduct materially inconsistent with the Charter. Anyone can report a concern to ASSUR.AI.
Reports are reviewed by a person. The member is normally given an opportunity to respond before its standing changes, except where immediate action is necessary to protect the integrity of ASSUR.AI.
Misuse of a badge may result in corrective action, suspension, removal from the Trust Register or rejection of renewal, depending on the seriousness and circumstances of the misuse.
Complaints about ASSUR.AI itself are handled separately from concerns about a member under the complaints route set out in the Membership Terms.