The record-keeping obligations for high-risk AI systems, in plain language. Four short pages.
Annex IV is the technical documentation a provider of a high-risk AI system has to compile and keep current, under Article 11(1). It covers how the system was built, how it was tested, how it performs, the changes made across its lifecycle, and how it is monitored once in use.
Article 12 sits alongside it. It requires a high-risk AI system to automatically record events (logs) over the lifetime of the system, so the functioning of the system stays traceable to a degree appropriate to its purpose.
The obligation is to show what the system did, traceably, after the fact. Not to show that a policy exists. When a decision is questioned, you need to reconstruct the events that led to it: the inputs, the steps, the version of the system in use, and the points where a person was involved.
Article 12 is explicit that the logs have to support traceability of the system's functioning, and Annex IV is explicit that documentation has to track the changes made to the system over time. Both point at the same thing: a record that holds up when read later by someone who was not in the room.
Most teams have a policy and a dashboard. Neither reconstructs what an agent actually did on a given run. A policy states intent. A dashboard shows aggregates. Article 12 asks for the run itself.
This is the gap Arkna fills: a replayable, hash-chained record of what an agent did, step by step, that a reviewer can verify independently. Arkna does not assess whether the decision was correct, and does not provide legal advice on the Regulation.
Regulation (EU) 2024/1689. Every claim on this page traces to the text below.