Adversarial Investigation Resource Exhaustion (AIRE): Degrading Agentic AI Defenses Via Token (and Related Capacity) Burning
4 minutes ago
1 min read
C/O Futures AI Initiatives Op-Ed
Kaden K. Bunker
19 September 2026
Artificial intelligence is rapidly moving from the sidelines of cybersecurity into the middle of the security operations center. Instead of merely summarizing alerts, AI systems are increasingly being asked to triage incidents, gather evidence, search logs, correlate activity across systems, consult threat intelligence, and recommend what defenders should do next. That promises faster investigations and less pressure on human analysts. It also creates a new question that security teams should start asking now: what happens when an attacker learns how to make the defender’s AI investigate too much?


Comments