Offsec Harness Index: an evidence index of AI penetration tools actually abused by attackers
Offsec Harness Index
offsecharnesses.ai's Offsec Harness Index includes 14 AI penetration and attack tools, item by item summarizing real abuse evidence from security vendor reports and labeling statuses such as CONFIRMED and OBSERVED. Among them, one control end of ARTEX attacked more than 200 organizations within 11 days, with 9 confirmed to have gained access permissions, and Storm-0940 used it to attempt logins to more than 2,250 Microsoft 365 tenants; Cairn was used in 25 operations targeting at least 19 organizations, and Gambit counted 105 related attack projects from September 10 to 15, with at least 27 companies breached. Attribution covers China-linked espionage, North Korean state actors, profit-driven crime and lone operators, with evidence from vendors including Zscaler, Check Point, Microsoft, Google and Anthropic.
The index summarizes reports from multiple security vendors, item by item recording the AI penetration tools attackers have actually used and the scale of intrusions, allowing abuse evidence to be checked by tool.