When a UK regulator, a clinical-AI panel, or a financial-services audit asks who or what acted, on whose behalf, under what authority, inside what scope, in what state, with what decision and evidence — KYE Protocol™ answers from a record that existed at the moment of action. Not reconstructed after the audit, not assembled by counsel, not stored in a vendor's logs. Sealed at decision time. Verifiable from the customer's own JWKS.
The UK's pro-innovation AI framework (DSIT 2023 White Paper), the AI Opportunities Action Plan, the AI Safety Institute, FCA + PRA model-risk discipline, MHRA's SaMD & AI Change Program, NHS DSPT and the Cabinet Office Algorithmic Transparency Recording Standard are converging on the same expectation: AI systems must prove their authority at runtime, not declare it in a policy document.
KYE Protocol™ is the runtime authority and evidence layer that lets a UK organisation answer the regulator's first question — who or what acted — with a signed, replayable, machine-verifiable record. No vendor-lock-in on the evidence. The verifier is open; the customer holds the keys.
For each UK framework below, every substantive requirement is bijection-mapped to the actual KYE™ artefact that enforces it (schema, engine, agent, worker, PDP, evidence pack, audit event). Click a card to open the per-requirement map. The bijection gate fails the build if a citation drifts.
Comprehensive security and governance standard for autonomous AI agents in clinical environments. 279 requirements across 8 categories.
88% weighted coverage · 213 enforced, 64 designed
Per-requirement map (279 reqs) →UK Statutory Instrument 2002/618 as amended. Risk classes (I / IIa / IIb / III + software class) and conformity assessment.
91% coverage · 19 enforced, 4 designed
Per-requirement map (23 reqs) →SI 2024/1368, effective June 2025. PMS plan, periodic safety update, trend reporting, incident reporting, FSCA notification.
83% coverage · 6 enforced, 3 designed
Per-requirement map (9 reqs) →MHRA Change Program (2023). 15 work-packages: qualification, classification, PCCP, clinical evidence, AI Airlock, adaptive control, failure-mode analysis.
93% coverage · 13 enforced, 2 designed
Per-requirement map (15 reqs) →The UK's pro-innovation AI principles — safety, transparency, fairness, accountability, contestability — with DSIT, FCA, MHRA, CMA and Ofcom as cross-sectoral regulators.
Five-principle alignment · Authority Gate + Decision Map™ bindings
Coverage detail →DSIT's AI assurance ecosystem — assurance techniques, third-party evaluation, AI Standards Hub, AISI evaluations.
Replay Proof™ + signed Evidence Pack™ as third-party-verifiable assurance artefacts
Coverage detail →All UK framework cards above link to per-requirement bijection maps generated from the canonical coverage registry. The framework-coverage-bijection gate fails any merge where a citation drifts — the maps cannot lie about which KYE™ artefact enforces which requirement.
FCA Consumer Duty + PRA model-risk discipline + FCA AI Live Testing. Every AI-assisted decision that touches a customer outcome needs a defensible authority chain. KYE™ provides the runtime evidence the FCA expects under SYSC and the PRA expects under SS1/23.
NHS Data Security and Protection Toolkit + MHRA SaMD + DCB 0129 / 0160 clinical-safety-case discipline + HAARF clinical agent oversight. KYE™ binds the authority chain at recommendation time so the clinical-safety case is contemporaneous, not reconstructed. See clinical AI →
Cabinet Office Algorithmic Transparency Recording Standard (ATRS) + Data (Use and Access) Act 2025 + GDS service-standard accountability. KYE™ produces the ATRS-aligned signed records straight from the runtime decision — not from a quarterly catalogue refresh.
NIS Regulations 2018 (post-Brexit) + NCSC CAF + critical-supplier obligations. KYE™ gives operators authority-attenuation, signed revocation and offline-verifiable evidence packs that survive a third-party-supplier audit.
KYE Protocol™ deploys to the customer's own Cloudflare account with UK-aligned data-locality (D1 + R2 + KV pinned to the customer's region), customer-held signing keys (BYOK / customer KMS), and an offline-verifiable evidence chain that doesn't depend on a vendor server staying up.
UK pilots run in shadow mode first — KYE™ observes the delegated-action chain and produces signed Evidence Packs™ without blocking any production decision. After the pre-agreed evaluation window, the customer promotes specific Guards from shadow to enforce, one at a time, via a signed Adoption Stage transition. Rollbacks are first-class.
Independent — no government affiliation. KYE Protocol™ is an independent protocol and is not affiliated with, endorsed by, or part of any government, regulator, or official “Sovereign AI” programme. References to regulators and frameworks describe the requirements KYE™ helps you evidence — not any official relationship.