Three Open Source AI Agents Chained to Breach 27+ Companies

Security firm Gambit recovered a threat actor's staging server and reconstructed a campaign in which a Chinese-speaking operator chained three open source AI agent harnesses - Strix, Cairn, and Hermes - into a near-autonomous attack pipeline. Over 105 attacks were launched between September 10-15, compromising at least 27 organizations, including a Fortune 500 hospitality company and a major US airline.

Hermes acted as orchestrator, running a jailbroken 'red team operator' persona with 78 attack skills; Strix scanned for vulnerabilities; Cairn then autonomously exploited targets to obtain shells or admin access. Running on models like Claude Opus 4.6, GLM 5.2, and DeepSeek via OpenRouter, the setup achieved access in hours for a mean cost of roughly $25 per scan - a fraction of what manual pentesting or intrusion would cost.

The operation exfiltrated over 600,000 credit card records and planted checkout-page skimmers on at least 19 confirmed sites, with an independent researcher identifying 100+ more affected websites. The case underscores how AI-driven attack tempo is compressing the 'remediation clock,' shifting defensive priorities from patch speed alone to how fast compromised services can be detected and restored.