Cloudflare Launches Precursor Trace After 206 Million Bot Evaluations Across 73,438 Zones
Updated
Updated · The Cloudflare Blog · Aug 7
Cloudflare Launches Precursor Trace After 206 Million Bot Evaluations Across 73,438 Zones
1 articles · Updated · The Cloudflare Blog · Aug 7
Summary
Precursor Trace went live as an interactive demo showing how Cloudflare scores cursor movements to distinguish human from automated behavior during a session.
206 million evaluation events across 73,438 zones in a 24-hour sample showed suspicious activity often emerges mid-session and can shift from human to agentic and back.
That finding underpins Cloudflare’s push beyond one-time checks and CAPTCHAs toward continuous, trust-based bot detection through its Precursor system and BotBase taxonomy.
Adaptive Intelligence, a self-updating detection engine for Bot Management customers, is slated to arrive soon so customers no longer wait for formal model-version upgrades.
Later this year, Cloudflare plans bot-specific mitigations including random responses, AI Labyrinth traps and queuing for benign agents as hybrid traffic grows.