Agent Docket

Where AI conduct meets the record

A tracked list of public legal matters in which the conduct of an AI system, or the record of what that system actually did, is at issue. Cases about autonomous actions and their consequences. Disputes over logs, audit trails, retention and spoliation. Matters where an institution's audit of its own systems was challenged and independent verification was demanded.

Every entry links to a primary source. Nothing here is a Hive claim. It is a reading list for the same question each of these matters eventually reaches: can anyone prove what the system actually did?

Loading the docket.

How this list is maintained

A scheduled job re-reads public court dockets and regulator publications and rewrites the feed at /assets/agent-docket.json. An entry appears only when the case name, court and date can be tied to a fetched primary source. Rows that cannot be verified are dropped rather than estimated, and if the feed itself cannot be loaded the ticker at the bottom of this site does not render at all. That rule is the point of the list.