How Scam Scanner detects scams

Every scan runs through the same sequence of checks, whether it starts from pasted text, a link, an email, a screenshot, a QR code, or a photo:

1. Read the content. The message is classified by type (a delivery text, a bank alert, an invoice, and so on). For camera captures, an on-device Vision classifier first works out what the image actually is before anything is analysed.
2. Check it against the brand-impersonation database. Around 93 brands relevant to Australians are recorded with their genuine domains and SMS sender IDs, so a lookalike domain or a sender a real brand would never use can be flagged.
3. Match known Australian scam patterns. The message is compared against the recurring tactics behind local scams — fake redelivery fees, unpaid-toll texts, “move your money to a safe account”, gift-card and cryptocurrency demands.
4. Analyse with AI. The content and those signals are passed to an AI model through a secure proxy, which weighs them together and explains its reasoning. The detection logic lives in the app; the proxy only forwards the request.
5. Return a score, the red flags, and next steps. The output is a 0–100 trust score, the specific warning signs found, and what to do next.

Rather than a blunt yes/no, Scam Scanner returns a trust score from 0 to 100 and sorts it into four verdicts — Safe (80–100), Caution (55–79), Suspicious (25–54) and Scam (0–24). This is deliberate: most real-world messages sit in a grey area, and a score with reasons lets you judge how much caution a specific message warrants instead of trusting a single label.

A high trust score means no obvious red flags were found — it is not a guarantee a message is genuine. The safest habit with anything involving money, logins or personal details is to verify directly with the organisation, however a check came out.

No automated scam checker is right every time, and we think it is more useful to be clear about that than to imply certainty. Scam Scanner can be wrong in both directions: a carefully crafted scam can score better than it should (a false negative), and a genuine but unusual message can be flagged as suspicious (a false positive).

So treat the result as informed decision-support, not a verdict. In particular:

• A low-risk result means “no obvious red flags found”, not “definitely safe”.
• Anything that asks for money, a payment, a login or a one-time code should be verified directly with the organisation through a channel you find yourself — never a link or number from the message.
• Scam Scanner is general information for Australians; it is not legal or financial advice.

The guidance on this site is grounded in Australian consumer-protection sources and a maintained brand database — not opinion. Our reporting advice and scam-pattern knowledge draw on the national bodies that track and act on scams:

• Scamwatch — Report the scam to the ACCC's national scam service.
• ReportCyber — Report cybercrime and financial loss to the police.
• ACMA — Report scam texts and spam SMS or calls.
• Forward to 7226 (SPAM) — Forward the scam SMS to short code 7226 so your carrier can block the source.
• IDCARE — Free identity and cyber support if your details were taken.

The brand-impersonation database records each brand's official domains and sender IDs from the brand's own published channels, so the comparison a scan makes is to verified facts rather than guesswork.

The cornerstone guides and scam-type explainers are written and reviewed by Reuben Schultz, the founder of Scam Scanner.

The per-brand “is this a scam?” pages are composed from that verified brand database, and no brand page is published until it passes a human review gate. Our governing rule is simple: we document only what we can verify — official domains, sender IDs, real reported scam patterns — and never invent brand facts, ratings or statistics to fill a page.

Scam tactics change constantly, so every guide and brand page carries a visible “last updated” date, and we revise the guidance when a brand's genuine details change or a new scam pattern starts circulating in Australia. If you spot something out of date or incorrect, tell us at support@tryscamscanner.com.