What PehleCheck checks

Messages and links

• Patterns used in Indian scams: fake KYC and account-block warnings, ‘digital arrest’ threats, task-based job offers, guaranteed-return investment pitches, lottery and refund lures, ‘new number’ messages from family.
• Hindi, English and Hinglish wording, including messages that mix scripts.
• Link structure: lookalike bank and brand names, misspelt domains, look-alike characters, URL shorteners, free hosting, and official Indian banking (.bank.in) and government (.gov.in) addresses.
• Domain age from the public registry (RDAP), and threat-intelligence lookups such as Google Web Risk, URLhaus and VirusTotal when they are configured.
• An isolated page check that follows redirects safely and reports login forms, payment requests and downloads, without your browser ever opening the page.

Screenshots, PDFs, QR codes and voice notes

• Text is read from screenshots on our own server (English and Hindi).
• QR codes are decoded. UPI payment QRs are explained in plain words, including who would receive the money and how much.
• PDFs are checked for embedded scripts and their text is analysed like a message.
• Voice notes and call recordings are transcribed, then checked, when a speech-to-text provider is set up.

Phone numbers and UPI IDs

• Whether the number is valid in the Indian numbering plan, and its type.
• Reports from other PehleCheck users, shown separately from verified listings. A report is a lead, not proof.
• UPI ID format and the handle’s provider, plus any verified listings.

An answer you can act on

• A risk level: Low, Suspicious, High, Critical, or Inconclusive when there isn’t enough evidence.
• The exact warning signs found, each labelled as observed, threat intelligence, a community report, a claim in the message, or AI interpretation.
• What we couldn’t check, and why.
• Practical next steps, things to avoid, and the official places to report.

Optional AI explanation

With your permission, an AI model reads the wording to explain the manipulation tactics in plain language. Card numbers, Aadhaar and PAN numbers, OTPs and account numbers are removed first. AI findings are shown as interpretation, are limited in how much they can move the score, and are discarded if they quote text that isn’t in your message.