ISO 42001 and the AI Act · For AI governance, compliance and legal teams
AI turns up in every team. The audit needs each one written down.
From your product documents, supplier contracts and evaluation reports, EA lists your AI systems and drafts what ISO 42001 and the European Union’s AI Act ask for: the policy, impact assessments and Statement of Applicability. Your lawyers set each system’s role under the Act, and your AI Governance Lead approves each piece.
Try EA free. No card needed.
Organisations that run work on EA
Most of the work is finding things out.
AI in places nobody listed
Product ships AI features, teams adopt AI tools, and suppliers build models into what you buy. Each one has to be found and given an owner before anyone can govern it.
Three teams per assessment
Data science knows the model, product knows the purpose, and legal decides the role under the Act. Each impact assessment needs all three in one document.
Dates set in law
Most of the AI Act has applied since 2 August 2026, and high-risk obligations begin on 2 December 2027. The certification audit brings a date of its own.
With EA keeping the inventory and the drafts, each team is asked only for its part, and the programme holds together between audits.
What EA does
Two jobs, run as one programme.
One year of the programme at a company that builds AI into its product and uses AI tools in-house.
01 / 06
- EA
Builds the inventory from product documentation, the sub-processor list and software purchase records: each system, its purpose, its model provider and its owner. Knowledge Library
Your teamOwners confirm their systems, and the AI Governance Lead sets what’s in scope.
- EA
Lays out each system’s facts for General Counsel beside the Act’s definitions of provider and deployer and its high-risk uses.
Your teamGeneral Counsel decides each role and risk class, and EA records the decision and who made it.
- EA
Runs the gap analysis against ISO 42001’s requirements and Annex A controls from the records you hand it, and lists the obligations under the Act that follow from each system’s role. Gap Analysis
Your teamThe AI Governance Lead confirms which controls apply.
- EA
Writes the AI policy, an impact assessment for each system and the Statement of Applicability (SoA), citing the runbook, evaluation report or contract each section came from.
Your teamThe Head of Product and the Head of Data Science fix anything that doesn’t match how the system really works.
- EA
Asks each owner for the evidence the obligations need, from each system’s model evaluation to the People Lead’s AI literacy training records, and follows up until it’s in. The Act has asked every provider and deployer to support their staff’s AI literacy since February 2025. It flags any evidence that no longer matches the system, like an evaluation run before the model changed. EA Agent
Your teamThe Head of Data Science reruns the evaluation on the current model, and the AI Governance Lead sees what is still due.
- EA
Gathers the year’s work for sign-off: the SoA and impact assessments for the AI Governance Lead to approve, the management review pack for the meeting your Chief Executive Officer chairs, and the audit pack for your certification body. Approvals
Your teamThe AI Governance Lead approves, with your Chief Executive Officer’s signed management review attached, and you share the pack with the certification body by a link you can revoke.
See it run on one of your own AI systems.
Book a demoAfter the audit
Each new AI system starts from the ones you’ve approved.
A new AI feature
EA adds it to the inventory, drafts its impact assessment from the closest approved one, and puts its facts in front of legal.
A model changes
Move to a new model or provider, and the impact assessments written for the old one are redrafted from the new contract and evaluation report. The Head of Data Science confirms each one describes the model you run today.
A role changes
When you put your own name on a system you bought, or change what it’s for, its role under the Act may change with it. EA sends General Counsel the new facts beside the old decision, and follows up until legal decides again.
A date comes up
Ahead of each surveillance audit and each date in the Act, EA reminds the AI Governance Lead of what is still due and who owes it, and asks the owners again.
- Assessment
- Impact assessment, support reply assistant
- Reviewed by
- Head of Data Science
- Approved by
- AI Governance Lead
- Date
- 22 October
What customers ask about your AI.
Customers’ security questionnaires now carry AI sections, and they ask about the systems in your inventory: what each one does, which provider runs it, and where a person reviews its output. EA drafts those answers from the inventory and assessments you’ve already approved.
EA’s own security
Where your AI records live, and who opens them.
Your inventory, impact assessments and legal’s memos are among the most sensitive records you keep. We ran our own ISO 27001 certification on EA: read how it ran on the ISO 27001 page.
- Hosted your way
EA’s cloud in a region you choose, hybrid options or on-premises. Data never used to train the models EA runs on.
- Access
Single sign-on, which you can enforce, and an access list for each workspace.
- Support access
EA staff see your data only in a time-limited window you open.
- Testing
An independent firm runs a penetration test on EA every year.
- Security contact
EA’s security team is at security@eaglobal.ai.
In the demo, see who can open legal’s memos, and who can’t.
Book a demoFAQ
Questions from governance and legal teams.
They’re separate: the Act is law, and ISO 42001 is a voluntary standard. A management system is a practical way to organise what the Act asks of you, which is why EA runs both as one programme.
An accredited certification body, working to ISO 42006, the 2025 standard for bodies that certify AI management systems. It audits your AI management system, issues the certificate and comes back each year for a surveillance audit; EA prepares the records it reads.
EA drafts from your records, and where none exists it marks the item “Needs evidence” rather than filling the gap. If an impact assessment calls for a person to review a system’s outputs and no procedure describes one, EA asks the Head of Product before anything is approved.
The people you add. Legal’s memos can sit in a workspace of their own, open only to General Counsel and the AI Governance Lead, and your certification body sees only the pack you share with it.
Your law firm keeps the judgement: roles, risk classes and what a contract promises. EA keeps the inventory, drafts from their positions and gathers the evidence, and brings them only the questions that need a lawyer.
Yes. An AI policy, a list of tools and a few assessments are enough to begin, and EA builds the inventory from there. Our Customer Success team helps with exports from other platforms. Plans differ by the number of autofills, the workspaces you need and the size of your knowledge library; see the pricing page.
Your next AI governance cycle, without a meeting for every system.
One AI system, to your AI Governance Lead’s approval.
Next to your AI programme.
Related solutions
- ISO 27001The two standards share a structure, so policies, internal audits and management reviews can run side by side.Security and IT
- Data protection impact assessmentsAn AI feature that uses personal data usually needs one alongside its AI impact assessment.Security and IT
- GDPR and data protectionModel providers sit on your sub-processor list as well as in your AI inventory.Security and IT
By industry
SoftwareAI features ship with each release, so the inventory changes as often as the product does.
Banking and building societiesCredit scoring is one of the Act’s high-risk uses, so legal’s classification carries more weight in a bank’s inventory.
Healthcare and life sciencesAI in clinical and diagnostic products can fall under the Act’s high-risk rules, so each system’s role is settled early.

