Security Tool

Phishing Explainer

Paste an email, SMS or link to understand phishing warning signs. Core analysis stays in your browser; optional indicator checks use remote services.

🔒 Core analysis runs in your browser. Optional indicator checks use remote services only when you start them; sharing full URLs requires separate explicit consent. This is an educational aid — it cannot catch every threat, so always use your own judgement.
349 approved regression tests Engine 1.4.0 Local deterministic analysis

The result is produced by versioned, reproducible rules. The engine distinguishes evidence that affects risk from informational evidence.

  • Regression baseline — Engine 1.4.0 is tied to an approved baseline of 349 automated tests.
  • Contextual detection — Rules evaluate whether gift-card, transfer or bank-detail language forms a current operational request and treat policies, training, quotations and completed notifications separately. They do not react to a single word alone.
  • RFC-aware email parsing — The local parser handles RFC 2047 B/Q encoded words, UTF-8, ISO-8859-1, Windows-1252 and US-ASCII, common RFC 5322-style address lists, folded or duplicate headers, and top-level versus quoted or nested provenance. Headers remain untrusted input: the tool does not certify their authenticity.
  • BEC correlation — bec-composite-v1 adds weight only when an executive or authority identity, a high-risk operational request and either secrecy or urgency with an operational pretext coexist. Isolated signals receive no composite-rule weight.
  • Privacy — The main deterministic analysis remains local in the browser. Optional indicator checks use remote services and start only when you request them: by default, the domain, registrable domain and SHA-256 hash of the URL are sent to the Cyberscan Worker. A full URL is sent only after separate explicit consent.
  • Versioned validation — The repository includes repeatable v2 and legacy benchmark suites. The v2 corpus uses synthetic, privacy-safe scenarios: it detects regressions but does not estimate real-world accuracy.
Explains the Why
Every red flag comes with a plain-language reason and the exact evidence from your text, so you learn to spot the next one yourself.
Local Analysis, Optional Checks
The message is analyzed in your browser. Optional checks send the disclosed indicators through the Cyberscan Worker to remote services; full URLs require separate explicit consent.

How to recognise phishing yourself

The anatomy of a phishing message

Almost every phishing attempt combines three ingredients: a trigger (urgency or fear), a disguise (a familiar brand or authority), and an action (click a link, enter data, pay). When you see all three together in an unexpected message, treat it as hostile until proven otherwise.

Read the domain right-to-left

The real destination of any link is the part just before the first single slash, read from right to left. In paypal.com.secure-login.ru/account the real domain is secure-login.ru — "paypal.com" is only a subdomain designed to reassure you. Checking the domain can reveal warning signs, but it does not identify every phishing attempt.

Legitimate companies never ask for secrets by message

Banks, postal services, social networks, and email providers will never ask for your password, OTP code, PIN, or card CVV via email or SMS. This rule has zero exceptions. Any message that does is phishing, no matter how official it looks.

Urgency is a manipulation tool

"Your account will be suspended in 24 hours", "act now", "final notice" — urgency exists to make you act before you think. Real organisations give reasonable notice and multiple contact channels. When a message rushes you, slow down.

Verify through a separate channel

If a message claims to be from your bank, don't use its links or phone numbers. Open the official app or type the known website address yourself. If it claims to be a person you know, contact them another way. A 30-second check beats a drained account.

What this tool can and cannot do

This Explainer detects known structural and linguistic patterns of phishing: deceptive domains, urgency wording, credential requests, brand mismatches, and more. It is deliberately transparent about its reasoning so you learn. But it cannot guarantee safety: a well-crafted attack may show few signals, and a legitimate-but-clumsy message may show some. Use it as one layer of judgement, never as the only one.

Related tools

To check a suspicious link directly, use the URL Scanner. To see if your email appeared in known breaches, use the Email Breach Check.