Sponsored by

eBook

The AI-Powered Adversary

Social engineering used to take time: researching a target, crafting a lure, waiting for a click. This ebook shows how attackers have removed that delay by pairing AI with the legitimate business tools people already trust. It draws on incidents the Huntress SOC investigated firsthand, including FakeAgent, a malvertising campaign in July 2026 that hit 29 organizations in two days. A sponsored search ad for a Claude desktop app led to a download page hosted on the real claude.ai domain, which quietly redirected to the SectopRAT infostealer, and Huntress has since tracked the same pattern across shared conversations on Claude, ChatGPT, Grok and other AI platforms. The ebook breaks down the ingredients of a believable lure, impersonation, urgency, familiar channels and borrowed infrastructure, explains why freely downloadable open-weight models without safety guardrails lower the bar for building malware, and argues that AI is not finding new ways in so much as exploiting the old gaps, exposed credentials, excessive permissions and unpatched systems, at machine speed.