All articles
Counterterrorism

How Do You Resolve Watchlist False Positives?

Lantern IntelligenceNovember 11, 20267 min read

Resolve watchlist false positives by treating a name match as a lead, not an identity finding. Compare independent identifiers, trace aliases to their original sources, check contradictory evidence, and document why the candidate is supported, excluded, or unresolved. Similar spelling alone does not establish that two records describe the same person.

What makes a watchlist match a false positive?

A watchlist false positive is a candidate match that incorrectly links the subject under review to a listed person. A matching name can trigger review, but identity resolution requires evidence beyond that initial similarity.

An unresolved match is different: available evidence does not yet support either linkage or exclusion. Keeping that distinction prevents uncertainty from becoming an unsupported finding.

In counterterrorism review, names may appear in account records, payment references, contact lists, travel-related documents, or messages. A name appearing in evidence does not establish that its owner controls an account, participated in a transaction, or engaged in wrongdoing.

Working definition: Identity resolution is the documented comparison of records to determine whether they refer to the same person, different people, or an identity that remains uncertain.

Why do aliases create watchlist false positives?

Aliases may include transliterations, nicknames, shortened names, reordered name components, former names, and account display names. Those categories carry different evidentiary weight.

A provider record associating a name with an account is not equivalent to a message mentioning that name. Neither automatically proves who operated the account at a particular time.

Common sources of confusion include:

  • Transliteration variants that make different names look similar.
  • Common names shared by unrelated people.
  • Handles reused across services or reassigned over time.
  • Shared phone numbers, addresses, devices, or network connections.
  • Incorrectly transcribed dates and identity-document numbers.
  • OCR errors in scans, photographs, or handwriting.
  • Repeated reports that all derive from one original source.

Alias rule: Preserve the original name exactly as recorded, and store any normalized search variant separately. Normalization should expand retrieval, not silently replace the evidence.

Which identifiers should investigators compare first?

Start with identifiers that can discriminate between candidates, while checking their source, reliability, and period of validity. No single identifier should be treated as universally decisive.

Evidence elementWhat it can help establishWhat can mislead reviewers
Full name and original scriptName structure and plausible variantsCommon names, transliteration, spelling errors
Date of birthA meaningful agreement or conflictPartial dates, placeholders, transcription errors
Document identifierA specific identity-record associationMisread characters, copied or misused documents
Provider account identifierContinuity within a serviceShared access or uncertain operator identity
Phone number or emailA contact or account associationReassignment, shared use, compromised access
Address or locationGeographic and temporal consistencyHousehold sharing, stale records, broad locations
IP or login recordAn access event associated with an accountShared networks, proxies, uncertain user attribution

Two matching fields are not necessarily two independent confirmations. If a spreadsheet and a report both copied the same source record, they provide repetition rather than independent corroboration.

Record where each identifier came from, when it applied, and what it actually establishes. Account ownership, account access, and personal identity are related questions, not interchangeable conclusions.

How can investigators resolve aliases step by step?

Use a repeatable review process that makes both supporting and contradictory evidence visible. The following is an analytical workflow, not legal advice or a substitute for organizational procedures.

1. What exactly triggered the candidate match?

Record the watchlist entry or reference, its version or retrieval date when available, the matched field, and the evidence item that triggered review. Preserve the original spelling and any available context.

Write a narrow question: “Does the account holder in this return correspond to the listed person?” Avoid starting with a conclusion such as “the listed person’s account.”

2. Which name variants are supported by evidence?

Create a working alias register. For each variant, note the exact text, original script if present, source location, relevant date, and relationship to the subject.

Separate observed aliases from search hypotheses. A nickname explicitly connected to an account in a source record deserves different treatment from a spelling variant generated for searching.

3. What independent identifiers support the link?

Compare dates of birth, account identifiers, contact details, document references, and other distinguishing attributes. Check the underlying records rather than relying only on summaries.

Ask whether each agreement is specific, independently sourced, and valid for the relevant period. A phone number associated with someone years later may not identify its user during the event under review.

4. What evidence argues against the link?

Look deliberately for contradictions: incompatible dates of birth, distinct document identifiers, different account-registration details, or records describing separate people.

Investigate the quality of the contradiction too. A date extracted incorrectly by OCR is not a reliable exclusion until checked against the original image. Missing data is not contradictory data.

5. Does the timing make the association plausible?

Compare registration, access, transaction, and communication dates. Distinguish the date an event occurred from the date a file was created, collected, or exported.

Record timezone assumptions and gaps. Apparent geographic conflicts may reflect network routing or shared access rather than impossible travel. Explain what the records support without overstating location precision.

6. What disposition does the evidence support?

Use clearly defined outcomes consistent with your team's procedures: supported linkage, excluded candidate, or unresolved candidate. Describe the basis rather than presenting a label alone.

For unresolved candidates, specify the missing discriminator and a focused next review task. Where operational consequences are significant, use an additional reviewer under your organization's process.

How should an investigator document a resolved match?

A useful resolution note allows another reviewer to reconstruct the reasoning without repeating the entire search. It should distinguish source observations from analyst interpretation.

Resolution checklist:

  1. Identify the candidate and the original match trigger.
  2. List observed aliases separately from hypothetical search variants.
  3. Cite source files and page, row, message, or record locations.
  4. Summarize supporting identifiers and their dates.
  5. Summarize contradictory evidence and how it was assessed.
  6. Identify source dependencies, gaps, and uncertainty.
  7. State the disposition, reviewer, date, and reopening conditions.

A hypothetical example: two records share a transliterated name, but the original documents contain different full birth dates and different identity-document numbers. That may support exclusion if the records are reliable and the differences cannot reasonably be attributed to recording errors. The name similarity should remain documented, alongside the reason it was insufficient.

How can Lantern support watchlist candidate review?

Lantern Intelligence is an AI evidence platform for investigators. It can support evidence review around a candidate match; investigators remain responsible for evaluating sources and deciding what an identity association means.

Lantern accepts any evidence file, including warrant returns from any provider, phone extractions, PDFs, scans, photos, video, spreadsheets, email, bank records, and call records. OCR turns scans, images, and handwriting into searchable text. Verify decisive extracted characters against the original evidence.

A practical review sequence is:

  1. Use Upload Case to bring the relevant evidence together.
  2. Use Search, Files, and Records to locate names and identifiers and inspect their context.
  3. Review relevant material through Persons, Messages, Media, and IP / Login Activity.
  4. Use Ask the evidence AI chat and AI Insights, including its intel brief and next steps, to assist review. Check AI-generated statements against source records.
  5. Use Export as part of preparing review materials, with your documented reasoning included in the review package.

The Dashboard is also live. Timeline and Relationship graph are coming soon, not current review tools.

Lantern runs in private AWS or on-prem on Google Gemma 4; evidence is never sent to an outside AI service. CJIS, FedRAMP, and SOC 2 are coming soon; Lantern does not currently claim those certifications or alignment.

See how it works and explore the use cases. For a related records workflow, read Terrorism Financing Investigation: Following Money Across Borders.

What should teams check before closing a candidate review?

Before closing, confirm that the disposition answers the identity question rather than merely restating a name match. Make uncertainty explicit, preserve contradictory evidence, and identify what new information would justify reopening the review.

The goal is not the highest number of matches. It is a defensible distinction between supported associations, excluded candidates, and questions that remain open.

Free sign-up includes 10 credits. Review the pricing page for current information.

Start free — get 10 credits

Put this to work on your own case

Sign up free and get 10 credits — enough to process a typical warrant return and see the messages, people, AI summary and recommended next steps.

watchlist false positiveswatchlist matchesalias resolutionidentity resolutioninvestigative evidence reviewcounterterrorism investigations
How it worksStart free