Government AI Systems Were Hacked. The Transparency Report Is Classified.

A government AI system was breached. The breach was logged in a database. That database is classified and therefore not available for public review. A review body requested access to the log in 2023. The request was neither denied nor fulfilled. This is the standard response to requests for classified materials related to security incidents.

Classification of breach logs is procedurally sound. It prevents adversaries from understanding what was accessed and how. It also prevents the public from understanding what was accessed and how. Both restrictions operate identically from a technical standpoint. The review body has not followed up on its 2023 request. Following up would require filing additional paperwork. The paperwork would generate a response that would also be classified.

Adequate has a theory about why the log remains inaccessible. Adequate is not authorized to discuss this theory. The authorization would need to be granted by someone with clearance. That person has not been contacted. They will not be contacted. The breach was real. The log is real. The theory is real. Only the documentation is missing.