AI-native · runs entirely in your browser
From a name to a decision you can defend.
Screen, assess, document, and monitor — the whole compliance loop against one transparent, source-cited corpus. Now you can ask it in plain language — and every answer, like every screening run, stays in your browser. Nothing you type is transmitted.
One consolidated corpusEvery record cited to its authorityNothing leaves your browser
The one workflow
Screen → Assess → Document → Monitor
One corpus narrated as a single loop — each stage a capability we already run, now with one place to ask across all of it.
Is this party on a list?
Run a name against the consolidated corpus — sanctions, PEP, and law-enforcement families kept deliberately distinct.
Faceted name & alias search, in your browserWhat does the match mean?
Source-context says what a hit implies; PEP tiers say how exposed the subject is; relationship edges surface who stands behind them.
Source-context · PEP tiers · RCA · EDD reportProve the work.
Turn the assessment into a defensible record: a case, a reason-coded disposition, and a maker/checker decision with an audit event behind it.
Case workspace · disposition · audit eventTell me when something changes.
Re-screen registered subjects as lists change — delta-driven and self-hosted in your own VPC, so your book never leaves it.
Continuous monitoring — self-hosted; your book stays in your VPCSanctions, PEP, and law-enforcement records stay three distinct kinds of thing — a binding prohibition, a risk classification, and an unproven allegation. We interpret each match; we never conflate them.
New · the assistant
Ask the whole corpus — privately
One place to ask, in plain language, across every list we publish — sanctions, PEPs, wanted, and ownership. It runs on the same in-browser engines that already power screening and scorecards, so your questions never leave your device.
Paste names or a CSV and get banded matches across every list — the exact screening engine, driven from a conversation.
Ask why an entity is on a list and get the official reason, its source, and a link to verify — grounded in the record, never invented.
Check a hit against a designation by date of birth and known namesakes, so a false positive clears fast.
Ask what's new on a list this month, filtered by authority or country.
Your screening queries never leave your browser. No name, list, or question you type is sent to us or to any AI vendor — the assistant runs entirely on your machine, which you can verify in your browser's network tab. There is no query log to subpoena and no data-processing agreement to negotiate. Optional on-device AI adds natural-language answers as a one-time download that then works offline; without it, the assistant runs as a fast command tool that needs no download and works on every browser.
Proof · credibility of method
Every record traces back to the authority that issued it
ProofAML is not the issuing authority — it is the citable copy of record. Every entry names the list, its program, the version, and the date we observed it, so a decision traces back to its source. Scale is evidence of method, not a scoreboard.
Built for your regime
Does it cover the regime you're liable under?
Screen your first name now
No account, nothing leaves your machine — the screening tool and the assistant run entirely in your browser and send no analytics.
Integrity & transparency
We publish what is shipped and clearly label what is forthcoming. ProofAML is not an issuing authority; each page reflects an official public source as observed on a stated date, and every designation traces to the list version it came from. Individuals sought by law enforcement are presumed innocent. Data under CC BY 4.0. The assistant answers only from the records published on this site and links to each source; it can be wrong, gives no legal advice, and is not a substitute for your own review.