Inside the Lab
Research in progress, open questions and observations — clearly distinguished from published findings.
Questions the lab is working on. Where one resolves into an evidenced finding, it moves into the published research.
What does an audit trail have to record before an agent's action can be called intended?
Transaction logging answers whether a step was authorised. It does not answer whether the sequence of steps served the objective the agent was given. The work follows what a behavioural log would need to capture to close that distance — and starts from the monitoring limits documented in The Monitoring Gap NIST Cannot Close.
Where does accumulated agent memory stop being context and start being an ungoverned control surface?
Memory persists across sessions and shapes later behaviour, while most governance boundaries are drawn around models and prompts. Follow-on work to What Your Agent Remembers — and Who Governs That.
How should authority be attributed when an agent acts through a human credential?
Identity systems were designed to answer who acted. Delegated, non-human identities operating with inherited permissions break that answer, and the attribution problem lands in the audit trail.
Questions under investigation, observations from operational environments, hypotheses being tested, and failure cases worth examining. Entries are working material: they state what is being looked at and why, not what has been established.
Conclusions, recommendations and findings belong in the published research, where the evidence and its limits are set out in full. Nothing on this page should be read as a settled result.
Published research →