How protection works
Known threats are checked against an active blocklist. New traffic is analyzed periodically or when configured thresholds trigger a scan. Findings are combined into confidence-weighted risk scores and feed future protection updates.
Signals and analysis
Checks cover bots, hosting networks, proxies, VPNs, repeated identifiers, abnormal click patterns, suspicious placements and device mismatches. Multiple AI analysis providers, including OpenAI, Anthropic and DeepSeek, can contribute findings; this does not imply vendor endorsement.
Limits and control
Core protection is included in platform access. No system guarantees complete fraud prevention. Set confidence thresholds, review evidence and override false positives with appropriate allowlists.
Related guides
Protection actions and confidence thresholds
Choose how the platform responds to network, device, supply and behavior risks.
Blocklists, allowlists and placement controls
Add manual controls without unnecessarily restricting useful inventory.
Traffic-quality reports for you and your clients
Explain delivered and filtered traffic with evidence under your own brand.