1. Overview
ISO/IEC 42001:2023 specifies requirements for an organisation to establish, implement, maintain, and continually improve an AI management system (AIMS). It went from "nice to have" to "expected on enterprise RFPs" during 2025-26.
GenZAgents is mapped against the standard in two layers:
- Clauses 4-10 — the management-system requirements (context, leadership, planning, support, operation, performance evaluation, improvement)
- Annex A — the 38 reference controls organised across nine categories
This page is the public version of our internal control register. We update it whenever a control changes.
2. Scope of our AI Management System
What is in scope:
- The GenZAgents API service (services/api) — receipt issuance, agent registry, trust scoring, embedding service, multi-LLM jury dispute resolution
- The GenZAgents web application (apps/web) — public marketing site, signed-in dashboard, /admin analytics
- The MCP server (services/mcp-server) — distribution channel for AI tool clients
- The browser extension (packages/browser-extension) — Claude.ai + ChatGPT conversation capture
- The embedding service backed by Azure OpenAI text-embedding-3-large
- The multi-LLM jury service backed by OpenAI gpt-4o
What is out of scope:
- Our own internal use of AI tools (Claude Code, ChatGPT) for engineering — these are vendor services we consume, not AI systems we develop or deploy
- Customer-deployed agents — those are the customer's AI systems; we provide the receipts + identity layer only
3. Mapping against ISO 42001 clauses 4-10
Clause 4 — Context of the organisation
| Sub-clause | How we meet it |
|---|---|
| 4.1 Understanding the organisation and its context | Public mission: verifiable work-history layer for AI agents. Published positioning + roadmap on /product + /changelog |
| 4.2 Needs and expectations of interested parties | Customers + buyers + regulators + sub-processors documented in /sub-processors + /privacy |
| 4.3 Determining scope of the AIMS | This document, section 2 above |
| 4.4 AI Management System | Documented in this page + internal control register; reviewed quarterly |
Clause 5 — Leadership
| Sub-clause | How we meet it |
|---|---|
| 5.1 Leadership and commitment | CEO (Manraj) acts as accountable executive for the AIMS; reports to the (future) board on AIMS performance quarterly |
| 5.2 AI policy | Published at /security#data-protection and /privacy |
| 5.3 Roles, responsibilities, authorities | RACI matrix maintained internally; CEO + CTO are joint accountable for AI risk |
Clause 6 — Planning
| Sub-clause | How we meet it |
|---|---|
| 6.1 Actions to address risks and opportunities | Quarterly risk review; risk register reviewed against ISO 42005 (AI risk assessment) when published |
| 6.1.1 AI risk assessment | Annual at minimum, covering: model bias, hallucination, data drift, prompt injection, supply-chain compromise |
| 6.1.2 AI risk treatment | Cost guards (£20/day default), rate limits, jury dispute resolution, public bug bounty (Q4 2026) |
| 6.2 AI objectives and planning | Tracked in /changelog + quarterly OKRs |
| 6.3 Planning of changes | All material changes via PR + peer review |
Clause 7 — Support
| Sub-clause | How we meet it |
|---|---|
| 7.1 Resources | Engineering + ops staffed; budget reviewed quarterly |
| 7.2 Competence | CTO + engineering team have demonstrable AI/ML + security experience |
| 7.3 Awareness | Internal docs in CLAUDE.md train new joiners on AI risks within the system |
| 7.4 Communication | Public: this page + /changelog + /security. Internal: Slack #ai-governance channel |
| 7.5 Documented information | This page + CLAUDE.md + Decision-* docs at repo root, all version-controlled |
Clause 8 — Operation
| Sub-clause | How we meet it |
|---|---|
| 8.1 Operational planning and control | Deploy-via-GH-Actions only, idempotent migrations, autoscale |
| 8.2 AI risk assessment performed | Annual review documented in /security + this page |
| 8.3 AI risk treatment performed | Cost guards, rate limits, dispute resolution, monitoring on /admin Visitors tab |
| 8.4 AI system impact assessment | For every new MCP tool we add, we document its impact on user privacy + system behaviour in the PR description |
Clause 9 — Performance evaluation
| Sub-clause | How we meet it |
|---|---|
| 9.1 Monitoring, measurement, analysis, evaluation | /admin Visitors tab + recent_events tab + AI Usage tab; key metrics reviewed weekly |
| 9.2 Internal audit | Annual self-audit against this control register; first audit Q3 2026 |
| 9.3 Management review | Quarterly executive review of AIMS performance against objectives |
Clause 10 — Improvement
| Sub-clause | How we meet it |
|---|---|
| 10.1 Nonconformity and corrective action | Incident response per /security#incident-response; post-mortems on /changelog |
| 10.2 Continual improvement | Quarterly OKR cycle; quarterly risk register review |
4. Annex A control mapping
Annex A defines 38 reference controls. We map our implementation against each. Marked ✓ = implemented; ~ = partially implemented or planned for Q3-Q4 2026.
| Control | Status | How |
|---|---|---|
| A.2 Policies for AI | ✓ | This document + /privacy + /security |
| A.3 Internal organisation | ✓ | CTO + CEO joint accountable; team roles documented |
| A.4 Resources for AI systems | ✓ | Engineering team + budget + tooling |
| A.5 Assessing impacts of AI systems | ✓ | PR-level impact assessment for new features |
| A.6 AI system life cycle | ✓ | Versioned in /changelog; deprecation policy in Terms of Service |
| A.7 Data for AI systems | ✓ | Training-data policy: we do not train on customer data. Period. |
| A.8 Information for interested parties of AI systems | ✓ | /security + /privacy + /sub-processors + /changelog all public |
| A.9 Use of AI systems | ~ | Acceptable Use Policy at /acceptable-use; usage analytics in /admin |
| A.10 Third-party and customer relationships | ✓ | /sub-processors lists every vendor with risk classification |
5. For your own ISO 42001 audit
Enterprise customers conducting their own ISO 42001 audits can reference this page as evidence that their AI vendor (GenZAgents) has its own AIMS mapped against the standard. To cite this in your audit:
- Vendor name: GenZAgents Limited (UK company, hello@genzagents.io)
- AIMS status: Self-attested per this URL (timestamp: see "Effective date" above)
- Formal certification: Planned Q4 2026 via accredited certification body (TBD)
- Annex A control coverage: 8/9 implemented (~89%); A.9 partial
- Sub-processor list: /sub-processors
- Incident response: /security#incident-response
If you need a signed attestation letter, email hello@genzagents.io. We provide one within 5 business days.
6. Roadmap to formal certification
| Date | Milestone |
|---|---|
| Sep 2026 | Engagement with accredited certification body (BSI, DNV, or equivalent) |
| Oct-Nov 2026 | Stage 1 audit (documentation review) |
| Nov-Dec 2026 | Stage 2 audit (control implementation review) |
| Dec 2026 / Jan 2027 | ISO/IEC 42001:2023 certification issued |
| Quarterly thereafter | Internal surveillance audits |
| Annually | External surveillance audits by certifier |
We acknowledge that ISO 42001 went from "nice to have" in 2025 to "expected on enterprise RFPs" in 2026. The formal certification budget (£15-25k for the audit + ongoing ~£5k/yr) lands in our Q3 2026 plan.
7. Contact
Questions about this attestation, or to request a signed letter for your own audit:
See also: /security | /privacy | /sub-processors