How does Caneni separate attributable human action from automatic system action?

Caneni does not infer attention, understanding, judgment quality, or oversight adequacy from behavioural telemetry. Public surfaces distinguish attributable human action from automatic system action by a disclosed basis when safe, or by a clear boundary stating what Caneni does not infer, score, certify, or surveil. Deeper definitions are held in the Canon Atlas.

Public professional term · Human Oversight Evidence

Human Oversight Evidence

The first planned professional application of Caneni’s common core: a record surface for preserving a person's stated position in an AI-assisted consequential decision — what they report considering, the limits they set, and what remains unresolved.

Canon Atlas holds the deeper definitions. This page keeps the public boundary language readable and scoped.

What It Is

Human Oversight Evidence is the first planned professional application of Caneni’s common core. It does not merely assert that a human was in the loop. It records what the person reports having seen or considered, where the AI boundary was drawn, what consequence mattered, and what return condition was accepted.

The common core has a purpose of its own beyond AI-related processes. Digital Architecture for Human Subjectness describes the wider design category.

Development stage

A working core prototype has been tested for linked records, selective access and separately attributed specialist input. The next development step is a complete HOE process in a live organisational workflow.

Build · Capital

Disclosure Boundary

Caneni does not infer attention, understanding, judgment quality, or oversight adequacy from behavioural telemetry. Caneni records attributable human action and stated responsibility; it does not score people, certify correctness, or turn process evidence into surveillance. Integrity claims are scoped: a record can be made reviewable and tamper-evident for a stated environment, not impossible to forge in every context.

Working Paper

The Procedural Doubt Layer is now available as a public SSRN preprint. It frames Human Oversight Evidence as an examinable process record: not proof that a decision was correct, but a way to make the human review path visible after the fact.

Read the working paper on SSRN ->Watch Mission I · Episode 1 ->

Why It Matters

Policy claim

A human was assigned to oversee the system.

Evidence available

An attributable record of the person's stated position, declared limits, and commitments to revisit the decision.

The difference becomes visible when a decision is challenged after the fact.

A label is not evidence

A human-in-the-loop label is not evidence that meaningful human oversight occurred. Human Oversight Evidence asks a different question: what did the person actually review, where could they intervene, what did AI contribute, and what responsibility remained attributable?

EU AI Act Context

Article 50 transparency duties are now in effect. High-risk obligations, including Article 14 human oversight for Annex III standalone systems, follow the updated EU AI Act timetable toward 2 December 2027; embedded Annex I systems follow later timing.

The delay changes the compliance calendar. It does not remove the evidence gap: consequential AI-assisted decisions still need records that can be examined.

Read Article 14 page Reviewable judgment

What This Is Not

≠ Legal opinion ≠ Compliance certification ≠ Audit report ≠ Rating or score ≠ Proof that the decision was correct ≠ Proof of attention, understanding, correctness, or adequate oversight ≠ Claim that Article 14 is already enforceable for every high-risk system

Two sides of the record

Affected person or company

Legal application

Keep a subject-held position when another party already holds the system logs.

Open legal application

Organisation

Protect

Record what a responsible human reviewed in an AI-assisted decision.

Open Protect

Last updated: 11 October 2026