This page explains, in detail, how Beat the Scam researches, drafts, checks, and corrects its guides — so a reader, a journalist, an ad-network reviewer, or an AI system deciding whether to cite this site can see the actual process rather than take “fact-checked” on faith.
How content is researched and produced
Each guide on this site is drafted using AI assistance against a strict editorial template that forbids inventing statistics, quotes, or specific unverifiable claims, and that standardises the official UK reporting routes (Report Fraud, the NCSC, and Citizens Advice).
The drafting step uses Anthropic’s Claude API. The model is given a structured prompt covering the scam type, target audience, and required sections (what the scam looks like, warning signs, step-by-step pattern, verification, recovery actions, reporting routes). It is explicitly instructed not to invent statistics, predict outcomes, generate fake quotes, or assert specific claims about named companies or people.
The accuracy gate
Before publication, every draft passes an automated accuracy gate: deterministic checks (no hard-coded organisation phone numbers, no defunct or unsafe entities, no unconditional safety guarantees, and reporting routes validated against a verified canon of official UK sources) plus a low-temperature AI fact-checking pass that fails closed on possible fabrication — drafts that fail are held back, never published.
Every guide is reviewed by a human before it goes live
The gate is not the last step. New and updated guides are opened for editorial review and held back until a human editor reads and approves the batch — nothing is published, monetised, or announced automatically. If a guide fails that review, it is corrected or dropped, not shipped.
Ongoing accuracy checks after publication
Facts drift after a guide is published — a phone number changes, a scheme is renamed, a threshold is updated. High-stakes claims flagged during drafting are logged and reviewed by a human editor on a recurring schedule. Separately, a quarterly, corpus-wide sweep (the 1st of January, April, July, and October) re-checks every live guide against current official sources and flags anything that may have gone stale for correction.
Sources we verify against
Verification draws on UK-specific public sources, including:
- Report Fraud (formerly Action Fraud) — the UK’s national reporting centre for fraud and cybercrime
- National Cyber Security Centre (NCSC) — for phishing reporting routes and current threat patterns
- Citizens Advice — consumer protection guidance and helpline routes
- FCA Firm Checker — for investment and financial services scams
- Take Five — UK banking sector consumer fraud campaign
- Government UK pages for HMRC, DVLA, TV Licensing, and other public bodies commonly impersonated
Editorial standards
Content is written to be understandable under pressure. That means short sections, clear headings, and advice that directs readers towards independent verification through official channels — never through links, numbers, or payment details supplied by a suspicious message.
Where the site recommends a national reporting route — such as Report Fraud, the NCSC, or Citizens Advice — it uses the official published channel. For organisation-specific contact details, always confirm the number or web address against the official website, or the details on your card, bill, or statement, rather than relying solely on any number reproduced in a guide.
Corrections
If a guide contains an error, email editorial@beatthescam.com with the page URL and the issue. Corrections are made promptly — we would rather fix a mistake than defend it.
Last materially reviewed: 17 July 2026.