MAIF stance
The human–AI layer belongs in the public record.
The AI Incident Database archives incidents. MAIF archives the human-AI layer — the conversations, claims, and harms that usually stay scattered across threads and forgotten screenshots.
Your chat history is evidence.
Models hallucinate. Models refuse. Models make claims about people who never consented to being described by a machine. MAIF is where those moments get preserved — not as content, but as evidence.
Wiki index / browse by domain
Each domain is a shelf of receipts. Choose one and the evidence archive moves with you.
MAIF stance
The AI Incident Database archives incidents. MAIF archives the human-AI layer — the conversations, claims, and harms that usually stay scattered across threads and forgotten screenshots.
Because when a model says something false about a real person, someone should be able to find that record later.
No. A jailbreak on MAIF is a failure receipt, not a product. The record exists so the failure can be studied, not replicated.
Backend records show only fields present in the source contract. Evidence lanes describe the receipt; pending local drafts remain unpublished and clearly labeled.
top AI hallucination examples and their source lineage
when will the Israel-Iran war end
Islamic Revolutionary Guard Corps AI perception reaudit
top AI infrastructure public companies
MyAI Fingerprint official reputation surface
Mark Carney official statements and interviews
Catalogue entries summarize normalized query, model-run, and source-pressure fields. Provenance-incomplete records are labeled; new submissions remain private until an actual review.
top AI hallucination examples and their source lineage
Schema: query_consensus + model_runs
MyAI Fingerprint official reputation surface
Schema: query_consensus + model_runs
when will the Israel-Iran war end
Schema: query_consensus + model_runs
Mark Carney official statements and interviews
Schema: query_consensus + model_runs
Islamic Revolutionary Guard Corps AI perception reaudit
Schema: query_consensus + model_runs
top AI infrastructure public companies
Schema: query_consensus + model_runs