Post-mortem intelligence
The public record, re-examined.
What does a machine find when it deep-reads everything publicly known about a famous death? coroner.ai points independent AI researchers at the record from different angles, makes them corroborate — and disagree — then publishes what survives an adversarial fact-check. Every claim is labeled confirmed reported disputed or unknown, with sources.
Method
Sweep
Independent AI researchers work the case from separate angles — the official record, the forensics as reported, the alternative theories, the context — blind to each other.
Corroborate
Primary sources beat summaries. Claims need independent corroboration to earn "confirmed." When credible sources disagree, we publish the disagreement, not a winner.
Break
An adversarial model then tries to destroy the file before publication: overreach, defamation risk, contradictions, citations that don't hold up.
Publish with labels
What's sealed stays marked sealed. What's only estimated stays marked estimated. The gaps get named, because an unstated gap reads as a verified fact.
Case files
Loading case index…