DIBS uses AI-powered fraud detection to protect every customer and provider on the platform — in real time.
Every device is fingerprinted on login. Banned devices can never re-enter the platform under a new identity.
All in-app messages are scanned in real time. Phone numbers, emails, and off-platform payment requests are blocked or redacted automatically.
Providers' GPS is tracked during active jobs. Jobs marked complete without the provider being physically present are flagged immediately.
Fake positive reviews, coordinated review bombing, and self-reviews are all detected using statistical analysis and account cross-matching.
Card velocity, new account high-value bookings, payout address changes, and chargeback patterns are all monitored in real time.
Every Tier 1 and Tier 2 alert triggers a Claude AI assessment that generates a risk score, fraud type classification, and recommended action.
• Banned device returns
• GPS never at job location
• Explicit off-platform payment
• Sex offender registry match
Account locked, payout held, CEO alerted, evidence package generated within 3 seconds.
• New account high-value booking
• Payout address change
• 2nd dispute in 90 days
• Review clustering pattern
15-minute transaction hold. Claude assesses risk. AUTO-RELEASE if LOW, human review if MEDIUM/HIGH.
• Same provider booked 6x/month
• GPS occasionally offline
• Dispute rate above 15%
• Slightly unusual review pattern
Account tagged for extra checks on all future transactions. Weekly automated review.
• Single VoIP phone number
• Single IP mismatch
• First-time minor issue
• Unusual booking time
Logged for pattern building. Auto-escalates to Tier 3 if same signal fires 3+ times.