Beyond the Human Horizon: Cogent Q4 2026 Threat Report
Cogent Research analyzed 43 million assets across 50+ Fortune 1000 companies to measure the attack paths humans and AI agents can exploit. The findings show AI-only paths forming three times as fast as human-viable ones and outlasting them, as policies rank their findings too low.
Key findings
Machine-viable attack paths surged 386% in a year and now outnumber human-viable ones three to one
26% of machine-viable attack paths contain no finding that the organization's own remediation policy ever requires fixing
83% of machine-viable attack paths persist for more than 30 days, compared with 55% of human-viable paths
Machine-viable paths run twice as deep as human-viable ones, with a median of 8 asset hops versus 4
64% of attack paths cross domains no single security tool observes, and the median critical path needs data from five tools
What's covered
The full analysis includes depth and per-hop yield comparisons for both path types, why severity-based remediation leaves machine-viable paths standing, and a methodology section detailing data sources and thresholds.


