Notes on evidence,
trust, and AI accountability.

The AI Evidence Gap

When questions arise about model behaviour, safety, compliance or accountability, customers, regulators and the public are often dependent on evidence controlled by the model provider itself.

Closing the AI Evidence Gap

A technical framework for verifiable AI interactions.

Most AI systems today rely on provider-controlled logs and post-hoc explanations — mechanisms that depend on trusting the organisation that generated them. This piece sets out the technical properties verifiable AI evidence requires, and a minimal receipt-based architecture that satisfies them.

Why AI Needs a Flight Recorder

An executive brief on the missing layer of AI accountability.

When an AI makes a decision that affects your business or your customers, you currently have no way to prove what actually happened. This brief explains the AI Evidence Gap, the flight recorder analogy, and the questions executives and regulators should be asking.

How to Read a Trust Receipt

A field guide for compliance officers, auditors, and investigators — no cryptography degree required.

A trust receipt is a small JSON document — typically 1–5 KB, depending on schema version, telemetry, and content mode — that records a single AI interaction. This article is the field guide for the moment when someone hands you one and asks what it says.