Corpus analyticsThe public record,
The public record,
measured honestly.
Understand the legal AI risk landscape in seconds. Explore trends, sources, attribution, and consequences, then inspect the records behind every signal.
Dataset snapshot
Inspect provenance before drawing conclusions.
The public corpus records when the source was checked separately from the latest tracked decision. Field coverage varies because court records and source documents do not use a common schema.
- Corpus checked
- Sep 10, 2026
- Latest tracked decision
- Sep 8, 2026
- Coverage window
- Apr 14, 2023 to Sep 8, 2026
- Public records
- 2,036
Public dataset snapshotChecked Sep 10, 2026 · Latest decision Sep 8, 2026 · 2,036 records · UTF-8
Data preview
Open searchable directory Inspect the published schema
DatePublic matterCourtImpactAmountSource
Completeness by field
Missing values remain visible; they are not imputed.
Source link2,032 100%
Court recorded2,035 100%
Decision date2,036 100%
Outcome text1,632 80%
Judge recorded54 3%
Named AI tool222 11%
Known monetary amount235 12%
Evidence-link tiers
Classification is based on the recorded URL host.
Publisher document archive1,882 92%
Secondary or other source105 5%
Docket or document mirror28 1%
Official court or government17 1%
Source link missing4 0%
Recorded human-review status
Review status is reported as recorded; missing status is not imputed.
Marked human-reviewed0 0%
Not marked human-reviewed2,036 100%
Source provenance
Read methodology Source-host concentration
Host counts describe where recorded links point, not source quality by themselves.
HostRecordsShare of corpus
Damiencharlotin.Com1,88292.4%
Reason.Com371.8%
Law.Justia.Com130.6%
Naturalandartificiallaw.Com130.6%
Caselaw.Findlaw.Com50.2%
Govinfo.Gov50.2%
Storage.Courtlistener.Com50.2%
Casemine.Com40.2%
Gao.Gov40.2%
Leagle.Com40.2%
Source Not Linked40.2%
Courtlistener.Com30.1%
Use the case directory to reproduce a filtered view and inspect every named source.
Open complete data table Field definitions and methodology