Beware Untrusted Simulators -- Reward-Free Backdoor Attacks in Reinforcement Learning

TL;DR

This paper introduces a novel reward-free backdoor attack method called Daze that exploits simulator dynamics to stealthily implant backdoors into reinforcement learning agents without needing reward access, demonstrating effectiveness across various domains and real hardware.

Contribution

The paper presents Daze, a new attack method that implants backdoors in RL agents without reward access, and provides formal proofs and empirical validation including real-world transfer.

Findings

Daze reliably implants backdoors in RL agents across multiple domains.

Backdoors can transfer from simulation to real robotic hardware.

The attack does not require observing or altering agent rewards.

Abstract

Simulated environments are a key piece in the success of Reinforcement Learning (RL), allowing practitioners and researchers to train decision making agents without running expensive experiments on real hardware. Simulators remain a security blind spot, however, enabling adversarial developers to alter the dynamics of their released simulators for malicious purposes. Therefore, in this work we highlight a novel threat, demonstrating how simulator dynamics can be exploited to stealthily implant action-level backdoors into RL agents. The backdoor then allows an adversary to reliably activate targeted actions in an agent upon observing a predefined ``trigger'', leading to potentially dangerous consequences. Traditional backdoor attacks are limited in their strong threat models, assuming the adversary has near full control over an agent's training pipeline, enabling them to both alter and observe agent's rewards. As these assumptions are infeasible to implement within a simulator, we propose a new attack ``Daze'' which is able to reliably and stealthily implant backdoors into RL agents trained for real world tasks without altering or even observing their rewards. We provide formal proof of Daze's effectiveness in guaranteeing attack success across general RL tasks along with extensive empirical evaluations on both discrete and continuous action space domains. We additionally provide the first example of RL backdoor attacks transferring to real, robotic hardware. These developments motivate further research into securing all components of the RL training pipeline to prevent malicious attacks.