Category

Governance for the AI Web.

The web is becoming a place where AI agents act on people's behalf. KYE Protocol™ governs the authority behind those actions — who authorised what, under what delegation, relying on which models and vendors — and proves it later. The category it defines is Governance for the AI Web™, and the risk it removes is Systemic Authority Risk™.

Governance for the AI Web, stage by stage

Demo data

You learn: How it works, one stage at a time. You can: Run it end to end.

  1. Secure browsing

    KYE Protocol™

  2. trusted browsing

    KYE Protocol™

  3. governed browsing

    KYE Protocol™

How this widget is built
Demonstrates
KYE™ runtime
Components
  • The stages this page describes
Flow
Each stage in order, as the page sets it out. Architecture diagram
Used on

1 · The category

Systemic Authority Risk™ is the accountability gap of the AI web.

Systemic Authority Risk™ is the inability to reconstruct who authorised what, under what delegation, relying on which models and vendors, across a shared AI ecosystem. It is distinct from systemic financial risk and from systemic model risk: the failure is not that a model is wrong, but that accountability itself cannot be reassembled after an AI-mediated action.

KYE Protocol™ reduces it by reusing the rails it already ships — it coins no new authority machinery:

  • Identity. Every actor — human, system, or agent — resolves to a canonical kye:<class>:<trust-domain>:<subclass>:<local> ID.
  • Discovery. Authority records and registries are projections of KYE™ Directory Search™ — one directory, many views.
  • Evidence. Every governed action carries a signed Evidence Pack™ under write-once retention, with Authority Finality™ guaranteeing the authority that stood at the moment of the action.
  • Delegation. Agents acting on a person's behalf are bound as first-class principals through the delegated-agent contract — the accountability evidence the EU AI Act and the NIST AI RMF expect you to produce.

2 · The arc

Secure browsing → trusted browsing → governed browsing.

Each era of the web added a layer of assurance. The padlock made browsing secure. Reputation and verification made it trusted. The agentic web needs the next layer: governance — provable authority and evidence behind every consequential action.

1

Secure browsing

1 TLS proved the channel was private and the server was who it claimed. It said nothing about authority.

2

Trusted browsing

2 Reputation, badges, and verification told a person whether a site was credible. It said nothing about who authorised an action.

3

Governed browsing

3 Governance for the AI Web™ resolves the authority behind every consequential action — human or agent — and makes it replay-provable from public evidence.

3 · The family

Four channels, one category.

Each member is a new public-facing product built over the existing KYE Protocol™ rails — they add no parallel authority machinery. One category, four surfaces of the same governance.

KYE™ Lens™

Web · consumer A privacy-by-design browser extension that overlays authority and evidence onto the web a person and their agents browse. Local processing, no user-browsing exfiltration, opt-in telemetry only. Learn more. Built on: kye: ID · KYE™ Directory Search™ · Evidence Pack™.

KYE™ Site Authority™

Web · publisher A site scan and badge that lets a publisher disclose a site's authority and evidence posture, with public lookup through the KYE™ Site Authority Registry™. The site UUID is the canonical kye: ID. Built on: KYE™ Directory Search™ projection · Evidence Pack™ · Authority Finality™.

KYE™ Build-Release Authority™

Software supply chain A GitHub App governing the authority, evidence, and provenance of software supply-chain changes — builds, dependency updates, and releases. The actor, human or agent, resolves to a canonical principal. Built on: Evidence Pack™ · write-once retention · actor binding.

KYE™ Production Action Authority™

DevOps · runtime Governs the authority of SRE, incident, and production actions — including an autoheal-partner integration acting on production. Each action proves who authorised it and on what basis. Built on: Evidence Pack™ · write-once retention · Authority Finality™.

4 · The red lines

Two non-negotiables hold across every surface.

No truth score.

KYE Protocol™ asserts only verifiable facts — that an action is evidenced, governed, disclosed, and that an authority exists for it. It never asserts that a claim, a page, or an output is "true" or "false". The honesty bar is definitional: the category proves the basis an action relied on, never the truth of the underlying content.

evidenced governed disclosed authority-exists

Privacy by design.

Processing is local by default. There is no user-browsing exfiltration. Telemetry is opt-in only. A person's web is theirs; the category governs authority over actions, not surveillance over people.

5 · Next

See the category in context.

Governance for the AI Web™ is the web-facing projection of the broader agentic-governance category KYE Protocol™ defines. Read the category page, or browse the directory and evidence rails the family is built on.

Agentic governance categoryDirectory SearchEvidence Pack™

Canonical KYE™ surfaces referenced on this page: KYE Protocol™.