Michael D. Dennis

About

Michael D. Dennis is a leading researcher in artificial intelligence, with a primary focus on the robustness and safety of deep reinforcement learning (RL) systems. His most influential work, the 2019 paper “Adversarial Policies: Attacking Deep Reinforcement Learning” (92 citations), fundamentally challenged the field’s assumptions about security in multi-agent environments. Dennis demonstrated that an adversarial agent could indirectly manipulate a victim’s observations through its own actions—a phenomenon he termed “adversarial policies”—revealing that RL policies are vulnerable even without direct access to another agent’s sensory input. This breakthrough reshaped how researchers think about AI safety in competitive and cooperative settings. Beyond this, his contributions extend to foundational work in multi-agent RL and the development of robust training methods. Dennis’s research has been widely recognized for its practical implications, influencing subsequent studies on adversarial robustness and policy security. His work continues to guide the design of safer, more resilient AI systems, making him a key voice in the ongoing effort to ensure that advanced RL agents behave reliably in complex, real-world environments.

Research Focus

Key Achievements

Top Papers

1. 1
   Adversarial Policies: Attacking Deep Reinforcement Learning

   92 citations · 2019

Key Collaborators

Contact & Links