Trust & Transparency Center

1. Data Modes & “Verified” Thresholds

To prevent spreading misinformation, ScamWatch explicitly labels all data sources so consumers are never misled:

• Verified Data: An alert or campaign is marked as “Verified” only when it has been confirmed by official law enforcement bulletins (such as the FTC, FBI, or state Attorney General alerts) or when multiple independent community reports confirm matching target infrastructure and templates.
• Demo Data: Seeded campaign records used to demonstrate platform capability during beta evaluation. These are simulated examples for educational testing.
• Live Data: Real-time alerts generated directly by incoming signals (such as live website check requests).

2. How Confidence Scores Are Calculated

We evaluate confidence based on factual markers rather than exaggerated claims. Our formula aggregates:

• Infrastructure Matches: Presence of domains or phone numbers listed on verified warning registries.
• Text Pattern Matching: Similarity to known smishing/phishing templates (such as SunPass toll demands or utility cutoffs).
• Report Density: The volume of independent, matching community reports logged within a short time frame.

3. False Positive, False Negative & Model Drift Warning

Machine learning models are subject to drift, evasion, and false detections. A “Low Risk” search result does not guarantee safety—it simply means our database does not currently possess evidence linking that indicator to known scams. Conversely, harmless links or newly registered domains might occasionally trigger caution warnings (false positives). Always exercise independent caution.

4. Human Review Policy

To prevent platform abuse and targeted harassment, public campaigns cannot be created automatically. Before a campaign is published to the public Florida Alerts Feed, a human moderator must review the reports, confirm redaction of all private identifiers (PII), and verify the fraud signatures.

5. Data Removal & Dispute Policy

If you represent a legitimate company whose domain, phone number, or brand name has been flagged in error, please contact us at abuse@scamwatch.org. We review all verification disputes within 24 hours and immediately white-list verified legitimate entities.

6. Capabilities & Limitations

7. Changelog