AI TRANSPARENCY

AI Assistance Disclosure

VERIQIA uses AI as a bounded editorial-assistance tool. AI output may be wrong, incomplete or misleading, so significant editorial and publication decisions remain under authorized human control.

Last updated 20 August 2026

Where AI may assist

AI may assist narrowly defined tasks such as entity and location normalization, ambiguous event clustering, structured analysis, proposed summaries, source comparison, proposed translations and channel-specific distribution copy.

Deterministic processing first

VERIQIA prefers deterministic methods for feed parsing, URL normalization, validation, exact duplicate detection, timestamps and other tasks that do not require generative interpretation.

Human editorial control

AI cannot mark verification complete, approve a Story Proposal, release a Story File, approve a Distribution Package or publish to an external channel. Those actions require an authenticated, authorized human.

High-risk material

High-risk or sensitive content requires stronger human review. AI output cannot remove a HOLD, resolve an open evidence request by assertion, or make high-risk content eligible for autonomous publication.

Limitations and correction

Machine-generated output may omit context, merge distinct events, misread a source or state uncertainty poorly. Editors review it against available evidence. If an error reaches publication, the Corrections Policy applies.

No artificial truth score

VERIQIA does not present an AI-generated truth percentage. Public trust status is drawn only from the approved vocabulary and remains distinct from In Focus prominence.