Find it, preserve it, then prove it.
Open-source material is only evidence once its origin is known. XS1 Intelligence builds models and tooling to discover material across the web, social platforms, commercial imagery and public records, preserve it with its provenance, and verify where, when and by whom it was made.
Frame under test
Checked before it is believed.
A fictional still from a video said to show a vessel aground in a harbor channel. Each check is a separate layer of evidence, and each is recorded before the video is used anywhere.
Copies found
Not yet traced
- Tue09:52Earliest copyTue 09:52: Earliest copy
- Tue10:14RepostTue 10:14: Repost
- Tue10:31RepostTue 10:31: Repost
- Tue11:02Repost, croppedTue 11:02: Repost, cropped
- Wed07:10New captionWed 07:10: New caption
Verification log
Five checks run on one still before the video is used anywhere.
What it covers
From first sighting to graded evidence.
Collection and verification is where most open-source work succeeds or fails. These are the parts XS1 Intelligence builds models and tooling for.
Discovery and collection
Models that find relevant material across the web, social platforms, commercial imagery and public records, against a collection plan built from the partner's requirements.
- Collection plan
- Multilingual
Preservation and chain of custody
Every item captured with its URL, time and a SHA-256 hash, and every later step written to a handling log, so its integrity can be demonstrated later.
- Hashing
- Custody log
Geolocation and chronolocation
Fixing where material was captured from terrain, skyline, signage and architecture, and when from shadows, sun position, weather and other dated events.
- Where
- When
Provenance and media forensics
Finding the earliest copy, reading what metadata survives, checking content credentials where they exist, and looking for signs of editing or synthetic generation.
- Earliest copy
- Synthetic media
Source grading
Source reliability and information credibility graded apart, so a usually reliable source with a doubtful report reads differently from an unknown source with a corroborated one.
- Admiralty Code
Corroboration
Independent sources counted once per origin. Reposts, syndicated copies and quotes of one report never add up to agreement.
- Independence
- Contradictions kept
Corroboration
Volume is not agreement.
The same fictional grounding, now weighed against other sources. Switch them on and off: independent origins raise confidence, reposts of one video count once, and a contradiction lowers confidence without disappearing.
Claim“A cargo vessel ran aground in the harbor channel on Tuesday morning.”
Sources
Corroboration graph
Assessed confidence
Moderate
Basis
- Two independent origins support the claim.
- The reposts trace back to the original video, so they count once.
Source types in use: 3 of 5
Confidence follows independent corroboration, not volume. A video reposted three times is still one source.
Source grading
Two grades, never one.
The Admiralty Code grades who is reporting and what is reported separately. The video above is F2: an account with no track record, saying something that fits what else is known.
Source reliability
- ACompletely reliable
- BUsually reliable
- CFairly reliable
- DNot usually reliable
- EUnreliable
- FReliability cannot be judgedThe video
Information credibility
- 1Confirmed by other sources
- 2Probably trueThe video
- 3Possibly true
- 4Doubtful
- 5Improbable
- 6Truth cannot be judged
Collection & verification
Material you need to stand behind?
Tell us what you need to establish and how it will be used. Every request is reviewed before work is scoped.