the always-on adversary:
AI agents never clock out
MSTIC (Judy Ng, Mark Parsons, Ben Koehl) will discuss how threat actors are leveraging open-source autonomous penetration-testing framework, to conduct end-to-end intrusions at AI speed with minimal human direction. These tools can orchestrate the full attack lifecycle, from reconnaissance and vulnerability research through initial access, privilege escalation, persistence, and data theft, which has reportedly been linked to attacks against more than 100 organizations across many sectors, worldwide. These tools can operate continuously and pursue multiple attack paths simultaneously, its activity may be noisy yet still overwhelm defenders, compressing the time from initial reconnaissance to privileged cloud access in under a day. The central takeaway is that agentic offensive tooling shifts the advantage toward adversaries through speed, scale, and persistence, requiring defenders to respond with rapid, fully scoped, and tightly coordinated containment and remediation.