Free · private · works offline
Not sure about a message?
Paste it below for a calm second opinion. Second Look explains the tricks scams use in plain words and tells you what to do next. It never opens links, and nothing leaves your device.
Something else to check?
What is the caller saying?
Tick anything that's happening. You can do this while they talk.
Check a QR code before you scan it
Stickers on parking meters, parcels, posters and "scan to receive money" requests can hide a fake website or a payment from you. Second Look reads the code on this device and shows what it would do. It never opens it.
Already tapped, replied or paid?
It happens to careful, smart people every day. Scams are built by professionals to work. What matters now is acting quickly and in the right order. Tell us what happened and we'll build your plan.
In danger or being threatened right now? Call your local emergency number. Second Look is a guide, not an emergency service.
So we can tell you which helplines to call. Saved on this device only.
Practice spotting scams
Ten made-up messages. Some are scams, some are real, and some you can't judge from the message alone. Learning to say "I'd check first" is the most useful habit of all.
How Second Look works
Built to be trusted: it's private by design, open about what it checks, and honest about what it can't do.
Your message never leaves this device
All checking happens in this browser tab. There is no server, no account, no AI service and no tracking. The page's security policy blocks it from making any outgoing data connection, even if we made a mistake in the code.
- Requests to other websites since this page opened
- 0
- Outgoing data connections allowed
connect-src 'none'- Works with no internet
- Checking…
Don't take our word for it: switch on airplane mode and keep using it, or open your browser's developer tools and watch the Network tab while you check a message.
What it looks for
Scams change their wording, but they rely on the same few levers: pressure, fear, greed, trust and secrecy, plus a way to get your money, codes or details. Second Look looks for those levers and for links that lie about where they go.
How well it works, measured honestly
We wrote labelled test messages (all synthetic) and measured the rules in rounds. Each round, a brand-new set of messages was scored once, before we looked at what it missed. The first round showed the rules were overfit to our own examples (only 36% of new scams caught), so we rebuilt them around the structure of scams rather than specific phrases.
The number we stand behind is the last round, scored once and never tuned: it catches about three in four new scams from the words alone, and wrongly flagged one of the 12 ordinary messages. Your answers to the three quick questions add what the words can't show.
What it can't do
- It can't tell you a message is safe. "No warning signs found" only means none of the common tricks appeared.
- It can't see who really sent a message, check phone numbers, or visit links. It reads the words only.
- It misses scams that start innocently, like a friendly "wrong number" chat that turns to money weeks later.
- It's English-only for now. Reporting help covers the US, UK, India, Canada and Australia, with general advice elsewhere.
- It reads text and QR codes, not screenshots or voice calls (you can type what a caller said).
- QR reading uses your browser's built-in scanner, which Chrome and Edge have on Android, Mac and ChromeOS. Elsewhere you can type the address your camera shows.
Credits
Built for Next Byte Hacks V5. No third-party libraries, trackers or datasets, and the two fonts (Atkinson Hyperlegible Next and Bricolage Grotesque, SIL Open Font License) are bundled with the app: every example message is synthetic and was written for this project. Advice is based on public consumer guidance from the US Federal Trade Commission and national cyber-security agencies. AI assistance (Claude) was used while writing code and copy, and every rule and message was reviewed by the team.