How Scam Shield works
One pasted message can trigger fifteen to twenty-five paid Telegraph miner calls, depending on how many links, emails, and IPs it contains. Every call is a real x402 micropayment on Base Sepolia — nothing here is mocked or cached.
Paste a message, email, or link.
The text is scanned locally for URLs, emails, and IP addresses — free, no miner calls yet.
Non-English text is translated first, so every downstream check runs on English.
The raw text always goes through the message-fraud, AI-text, and prompt-injection miners.
Every URL found is checked for link safety and certificate trust, and its page content is extracted.
If a URL's extracted page reads as substantial, it's re-run through the fraud and AI-text jury too — catching scam landing pages even when the message itself looks clean.
Every email and IP found gets its own reputation and geolocation checks.
Results are rolled into one verdict, and every finding — cost, latency, signal hash included — is published to the ledger.
Why miners overlap on purpose
Several categories above are covered by more than one miner deliberately. Link safety alone is checked by six independent services; certificate trust by three. If one is slow, wrong, or offline, the scan doesn’t stop or wait for it — the others still answer, and the verdict is built from whichever miners actually responded. No single miner going down, or being wrong, can take the whole check with it.
Message-fraud classifiers — always run on the text
Classifies scam-message patterns — credential requests, investment pitches, payment pressure.
$0.01Checks against documented fraud schemes and known scam wallets.
$0.01Answers whether a message matches a known fraud pattern, with sources.
$0.01Fact-checks specific claims in the text against live evidence.
$0.01Judges whether there's enough evidence to act on this message at all.
$0.01AI-generated-text jury — always run
Detects AI-generated text via a Bittensor subnet model.
$0.02Fine-tuned classifier for AI-generated vs. human-written text.
$0.01Stylometry ensemble estimating the probability text is AI-written.
$0.01Prompt-injection detection — always run
Scans for indirect prompt-injection attempts hidden in the text.
$0.01Link safety — checked per URL found
Checks a URL against a crowd-sourced malware and phishing feed.
$0.01Checks a URL against the community phishing database.
$0.01Scores a domain against 70+ antivirus and threat-intel engines.
$0.01Aggregates phishing, malware, and domain-age signals for a URL.
$0.01Live reachability, HTTPS, and redirect check for a URL.
$0.01Sandboxed render of a page, capturing verdicts and redirects.
$0.01Certificate trust — checked per URL's domain
Validates a domain's live TLS certificate.
$0.01Looks up the most recent certificate-transparency record for a domain.
$0.01Deep TLS configuration grade — A+ through F — for a host.
$0.01Content extraction — checked per URL
Pulls the title and body text off a linked page.
$0.01Sender forensics — checked per email or IP found
Reputation lookup for an email address — leaks, blacklists, abuse.
$0.01Geolocation and privacy/abuse risk flags for an IP address.
$0.01Geolocation for an IP address from two independent providers.
$0.01Translation — runs first for non-English input
Translates non-English input to English before analysis.
$0.01