Publications

Two papers — one on AI-driven cybersecurity threats, one on generative agents simulating human behavior — read independently, describing the same architecture. The adversarial attacker and the generative agent run identical memory-reflect-plan loops. What differs is only the objective function.