This page exists because we believe compliance professionals deserve a precise account of what AML Checker can and cannot do. Overstating a tool's capabilities in a compliance context is not just bad marketing - it creates real regulatory risk for the firms that rely on it.
AML Checker runs structured OSINT queries across 9 signal profiles using real-time web search powered by Claude (Anthropic). Every result is scored by source authority and publication recency. The tool returns a structured risk band (LOW / MEDIUM / HIGH), a source-cited compliance narrative, and a tabbed signal breakdown showing results by profile.
The SAR tier additionally accepts uploaded identity and financial documents, extracts data from them using Claude's vision capabilities, and generates a structured SAR narrative with risk factors, evidence citations, and an FCA-aligned format - exported as a PDF.
AML Checker uses live web search to find publicly indexed references to OFAC, SDN, FCA, EU/UN sanctions lists, NCA enforcement actions, and equivalent sources. It is not connected to those databases via a direct data feed. This means it will find a Reuters article reporting an OFAC designation, a government press release about a conviction, or an FCA enforcement notice - but only if that information has been publicly indexed. A very recent designation or an enforcement action not yet reported publicly may not appear.
This is meaningfully different from tools like Dow Jones Risk & Compliance, World-Check, or ComplyAdvantage, which receive direct structured feeds from sanctions list publishers. AML Checker does not replace those tools for firms with a real-time screening obligation.
A LOW risk result means no significant adverse signals were found in publicly indexed sources at the time of the search. It is not a clean bill of health. The absence of results reflects the limits of what is publicly available and indexed - not a definitive absence of risk.
Searching a name like "John Smith" will produce high volumes of irrelevant results and low signal quality. AML Checker performs best with distinctive names, and accuracy improves significantly when additional identifiers are included - such as a company name, jurisdiction, or role. For common names, results should be treated with particular caution and cross-referenced independently.
SAR outputs are structured drafts designed to assist your firm's MLRO or nominated officer. Filing a SAR with the National Crime Agency is a legal act that must be performed by your firm's nominated officer. AML Checker produces a draft - it does not and cannot submit anything to the NCA on your behalf.
In the current version, the SAR is exported as an HTML file that you open in your browser and print to PDF. There is no server-side PDF rendering. The formatting is designed for print but the output depends on your browser's print settings.
Nothing produced by AML Checker constitutes legal advice, regulated financial advice, or a compliance determination. If you are uncertain about your obligations under MLR 2017, POCA 2002, or the Terrorism Act 2000, consult a qualified legal adviser.
| Limitation | Detail |
|---|---|
| Scan time | 15-25 seconds per scan - inherent to running live web searches within a single API call. This cannot be materially reduced without architectural changes |
| Common names | Ambiguous or common names produce lower signal quality and higher false-positive rates. Results require closer manual review |
| Date accuracy | Publication dates are not always returned by the underlying search API. When unavailable, a 24-month age estimate is applied for freshness scoring |
| Deep scan false positives | The optional expanded profile uses a broad OR-chain query and will surface unrelated results for common names. It always requires manual analyst review and is labelled accordingly in the results |
| No stored audit trail | Scan results and SAR outputs are not stored by Parrot Systems Ltd. You are responsible for downloading and retaining your own outputs for compliance record-keeping purposes |
| Single device per session | Your access token is stored in your browser's local storage. You will need to re-enter it on a different device or browser |
| PDF export format | SAR exported via browser print-to-PDF. Server-side PDF rendering is on the roadmap for a future release |
| Source dependency | Results depend on what is publicly indexed at the time of the scan. Recently published enforcement actions, new designations, or sources behind paywalls may not appear |
AML Checker is designed to support compliance workflows under the Money Laundering, Terrorist Financing and Transfer of Funds (Information on the Payer) Regulations 2017 (MLR 2017). Parrot Systems Ltd is not an FCA-regulated firm and AML Checker is not itself a regulated service.
Use of AML Checker does not satisfy, substitute for, or guarantee compliance with any regulatory obligation. Your firm's AML compliance programme, risk appetite, policies, and procedures remain entirely your own responsibility.
The SAR narrative output is designed to align with the NCA's guidance on SAR structure and content. It is a drafting aid only. Your MLRO must independently review, verify, and take full responsibility for any SAR submitted to the NCA.
AML Checker uses Claude (developed by Anthropic) to conduct searches and generate compliance narratives. AI-generated content can contain errors, misattributions, or inaccuracies - particularly when source material is ambiguous or when a searched name is common.
Every narrative must be reviewed against the source links returned alongside it. Do not rely on narrative text alone. If a cited source does not support the claim made in the narrative, disregard that element and investigate independently before taking any action.
To prevent misuse, AML Checker applies the following per-IP rate limits:
If you are running high-volume screening and require a higher limit, contact us at hello@aml.business.
If you have questions about the scope or limitations of AML Checker before relying on any output for a compliance decision, contact us at hello@aml.business.