iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
ScamShield is an AI-assisted scam triage project built around a specific need: helping its creator’s grandmother assess suspicious messages and know what to do next. Its author describes tools for checking text, URLs and screenshots, answering questions in plain language, and guiding people through recovery steps. But ScamShield is a project, not an independently validated security product: its older-adult user had not tested the interface, and its reported test results cover only four synthetic examples.
What ScamShield is designed to do
Farhan describes ScamShield as a browser client backed by a Flask API, with Google’s Gemma API for hosted use and an option to run inference locally with Ollama. The intended user is his paternal grandmother, or “dadi,” who uses WhatsApp and SMS and might ask, “Is this real, or is someone trying to trick me?” These are the author’s project descriptions, not findings from an independent inspection or test. Read the project article.
Scam checks and conversation
The author lists a multimodal scanner for screenshots and other visual material, plus inspection of message text and phishing URLs. A conversational assistant is intended to let someone ask about a situation in ordinary language—for example, what to do if a caller claiming to be police says a parcel contains drugs.
Recommended Free Tools
Guidance and additional features
A recovery wizard is described as pointing users to reporting steps and helping draft a complaint. The project article also lists family alerts through Discord or WhatsApp, a quiz, a threat-intelligence feed, a security hygiene scorecard, phone carrier lookup and crowdsourced reporting. The article does not establish that these features have been independently assessed or that alerts, scores or scam classifications are dependable in real incidents.
#1 Best Overall
Hosted and local use: what the author says
The author says the hosted demo sends requests to Google’s API, while local mode processes images on the user’s machine. Those statements describe the intended architecture; the application and its data flows were not independently audited. The article does not establish that all information remains local in every configuration, or document comparative hardware requirements or a complete privacy review.
| Consideration | Hosted demo, as described by the author | Local mode, as described by the author |
|---|---|---|
| Where image processing happens | Requests are sent to Google’s API, according to the author. | Images are processed on the user’s machine, according to the author. |
| Internet and API access | The described route uses Google’s API; connection and access are therefore needed for that route. | Not fully specified in the project article; the article does not establish whether every local workflow works offline. |
| Hardware and setup demands | Not stated in the project article. | Not stated in the project article. |
| What has been tested | The author reports synthetic-example tests; no independent evaluation of either deployment path is established. | The author reports synthetic-example tests; no independent evaluation of either deployment path is established. |
Because these are author-reported descriptions rather than an audit, do not treat either mode as a verified privacy guarantee. Avoid entering sensitive information unless you understand how the specific version you are using handles it.
What the reported test results do—and don’t—show
Farhan reports 54 passing automated tests, with a runtime of 27.22 seconds. For four built-in synthetic examples—two labeled safe and two labeled scams—he reports 100% agreement and a median latency of 8.07 seconds. These figures describe the author’s tests on a tiny, constructed set. They are not a real-world detection rate, an independent benchmark, or evidence that ScamShield will catch live scams or correctly classify a message a person receives.
Was ScamShield tested with its intended older user?
No. The author explicitly says his grandmother had not tested the interface and that there were no real user quotes to share; a sit-down handover was planned. The design persona explains the project’s motivation, but it does not demonstrate that an older adult can use the interface easily, understands its advice, or benefits from it in practice.
What to do if an online financial scam happens in India
If money has been lost or a fraudulent transaction is underway, report it promptly through official channels rather than relying on an AI tool to recover funds. India’s National Cyber Crime Reporting Portal directs people to call 1930 for immediate reporting of cyber financial fraud and also accepts reports online. Open the National Cyber Crime Reporting Portal.
- Call 1930 promptly to report online financial fraud. The portal’s guidance describes a follow-up complaint process after the call; follow the instructions given to you.
- Keep relevant information ready. The official instructions identify details such as the bank or wallet involved and transaction information. Read the official citizen reporting instructions.
- Use the portal’s online route if appropriate, and retain the details of your report and any follow-up instructions.
Prompt reporting is the supported advice; neither the portal instructions nor the project article establishes that reporting guarantees a fund recovery or a transaction freeze.
Rank #4
- Used Book in Good Condition
How to treat ScamShield’s advice
- Use it, at most, as a triage aid—not as proof that a message is safe or fraudulent.
- Do not act on a suspicious message’s payment demand, link, or request for credentials just because a tool labels it safe.
- Verify urgent claims through a contact channel you find independently, rather than replying to the sender or caller.
- If financial fraud has occurred in India, report promptly through the official 1930 and portal process.
ScamShield’s concept is aimed at making scam guidance more approachable for a family member, and its author reports a small set of passing tests. Its effectiveness, privacy properties and usability for older adults remain unverified by the evidence described in the project article.
Free tools Windows power users keep installed
One-click scans. No signup required.
Quick Recap
Best Value
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
