AI Fraud Risk Scanner
Detect fraud signals, social engineering patterns and AI assisted scam indicators before they cause damage.
Rolling activity model | refreshed locally
Verified reference: the FTC recorded 2.6 million fraud reports and 6.47 million total consumer reports in 2024. Public FTC data is periodic, not a live global AI-fraud feed. View FTC data
Awaiting analysis.
Risk categories
Why this content is risky
Fraud attack chain
Likely fraud types
Detected indicators
Evidence highlighting
Recommended actions
Report
AI Fraud Risk Questionnaire
Use these questions when you have a suspicious call, video meeting, message or request but no text to paste into the scanner.
Fraud categories this scanner covers
Each category below lists the signals the rule engine looks for. Select the matching content type in the scanner above to weight these signals correctly.
Global AI Fraud Watch
A monthly editorial update area for notable AI-assisted fraud patterns around the world. This section is intentionally manual so you can review and approve each update before publishing it.
Last updated
August 2026
Update cadence
Monthly, reviewed manually
Data status
Editorial summary – no live feed
| Pattern to watch | Typical scenario | What to verify |
|---|---|---|
| Voice-cloned executive calls | A familiar voice requests an urgent transfer, gift cards or secrecy. | Call back using a known number and require a second approval. |
| Deepfake video meetings | A synthetic face or scripted meeting pushes a payment or data request. | Confirm the meeting independently and avoid changing payment controls. |
| AI-polished recruitment scams | A convincing recruiter offers remote work, then asks for fees or identity documents. | Verify the employer domain, job listing and interview contact separately. |
| Personalised phishing at scale | A polished message uses current events, local language or leaked details. | Open the official app directly; never use the supplied link or QR code. |
Suggested monthly sources to review before updating this panel: national cyber-safety advisories, banking fraud alerts, law-enforcement notices and reputable incident-response reports. Keep claims linked to a dated source when you publish an update.
Fraud knowledge base
What is AI fraud?
AI fraud refers to scams and social engineering attacks that use artificial intelligence tools to write, translate, personalise or scale deceptive messages. This includes phishing emails drafted by language models, cloned voices used in phone scams, and chatbots that impersonate customer support or romantic partners. The underlying goal is the same as traditional fraud – obtaining money, credentials or access – but AI can make the delivery faster, more polished and harder to distinguish from genuine communication.
How is AI used in fraud?
Generative AI is commonly used to draft grammatically flawless phishing emails in multiple languages, generate fake job postings and interview scripts, produce realistic-sounding investment pitches, create synthetic profile photos for romance scams, and automate replies at scale so a scammer can run many conversations simultaneously. AI does not create new categories of fraud so much as it removes the friction – spelling errors, slow response times, language barriers – that used to make scams easier to spot.
How does this fraud scanner work?
The scanner runs a transparent, rule based engine entirely inside your browser. It scans the text you provide for weighted indicators across categories such as urgency, financial requests, credential harvesting, impersonation, threats, secrecy and guaranteed-return language, then combines the matches into category scores and an overall score. A separate set of linguistic heuristics estimates how consistent the writing is with AI assisted or scripted fraud content. Nothing is inferred by a trained machine learning model, and no content is transmitted anywhere.
What fraud signals does it detect?
The engine looks for urgency and pressure language, unusual payment requests, requests for passwords or one-time codes, impersonation of authority figures or brands, threats of account suspension or legal action, requests for secrecy, guaranteed-return investment claims, job-scam fee requests, romance-scam emotional patterns, cryptocurrency wallet and seed-phrase requests, and structural characteristics of AI generated persuasive text. It can also inspect a URL’s structure locally for common phishing indicators.
How to identify AI assisted scams
AI assisted scam content often reads as unusually polished for the context, mixes generic personalisation (“Dear valued customer”) with specific pressure tactics, repeats persuasive structures across paragraphs, and maintains an oddly consistent tone even when describing an emergency. None of these traits alone prove AI involvement – a careful human scammer can write the same way – but the combination is a useful signal to slow down and verify independently.
How to protect yourself from AI fraud
Verify unexpected requests through a separate, independently confirmed channel rather than replying to the message itself. Never share one-time passcodes, verification codes or recovery phrases with anyone who contacts you. Treat urgency and secrecy as warning signs rather than reasons to act quickly. Confirm bank account or payment detail changes by phone using a number you already trust, not one supplied in the message. When in doubt, pause and ask a second person before sending money or credentials.
AI fraud detection limitations
This tool cannot confirm whether text was written by a human or an AI system, cannot verify anyone’s real identity, cannot check whether a bank account, wallet address or company is legitimate, and cannot access, visit or scan the destination of a URL. Its scores describe how closely content matches known fraud patterns, not a certainty of fraud. Genuine messages can occasionally trigger indicators, and sophisticated fraud can occasionally avoid them. Always use independent verification before acting on financial or credential related requests.
Frequently asked questions
About this tool
AI Fraud Risk Scanner is a client side, rule based assessment tool. It runs entirely in your browser using JavaScript pattern matching against a transparent, weighted indicator list – there is no trained machine learning model, no external AI API, and no server. Scores reflect how strongly submitted content matches patterns commonly seen in email fraud, phishing, business email compromise, investment and job scams, romance scams, cryptocurrency fraud and AI assisted social engineering.
Because the tool has no access to real-world identity, account or transaction data, it cannot confirm that any specific message, person, website or transaction is fraudulent. Use it as one input among several – official verification channels, bank fraud teams and law enforcement remain the authoritative sources for confirming fraud.