mlsanityData to 14 Oct 2026

Angel or Demon: Investigating the Plasticity Interventions' Impact on Backdoor Threats in Deep Reinforcement Learning

arXiv 2605.14587 ↗First posted 14 May 2026cs.LG cs.AI cs.CRcited by 1

On the map: Deep reinforcement learning methods → Adversarial robustness and RL in finance

Abstract (arXiv)

Extensive research has highlighted the severe threats posed by backdoor attacks to deep reinforcement learning (DRL). However, prior studies primarily focus on vanilla scenarios, while plasticity interventions have emerged as indispensable built-in components of modern DRL agents. Despite their effectiveness in mitigating plasticity loss, the impact of these interventions on DRL backdoor vulnerabilities remains underexplored, and this lack of systematic investigation poses risks in practical DRL deployments. To bridge this gap, we empirically study 14,664 cases integrating representative interventions and attack scenarios. We find that only one intervention (i.e., SAM) exacerbates backdoor threats, while other interventions mitigate them. Pathological analysis identifies that the exacerbation is attributed to backdoor gradient amplification, while the mitigation stems from activation pathway disruption and representation space compression. From these findings, we derive two novel insights: (1) a conceptual framework SCC for robust backdoor injection that deconstructs the mechanistic interplay between interventions and backdoors in DRL, and (2) abnormal loss landscape sharpness as a key indicator for DRL backdoor detection.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Adversarial robustness and RL in finance, one of 15 directions in Deep reinforcement learning methods. Builds on 56 of its 74 references that are in the map. First posted on arXiv 14 May 2026.

Builds on

56 of 74 references.

Closest ancestors · 6

Same direction, newest first

  1. 2025TooBadRL: Trigger Optimization to Boost Effectiveness of Backdoor Attacks on Deep Reinforcement Learning11 Jun
  2. 2025Fox in the Henhouse: Supply-Chain Backdoor Attacks Against Reinforcement Learning26 May
  3. 2025UNIDOOR: A Universal Framework for Action-Level Backdoor Attacks in Deep Reinforcement Learning26 Jan
  4. 2024Online Poisoning Attack Against Reinforcement Learning under Black-box Environments1 Dec
  5. 2024Adversarial Inception Backdoor Attacks against Reinforcement Learning17 Oct
  6. 2024SleeperNets: Universal Backdoor Poisoning Attacks Against Reinforcement Learning Agents30 May

Other references · 50

  1. 2017Proximal Policy Optimization Algorithmscited by 15,215
  2. 2015Continuous control with deep reinforcement learningcited by 3,407
  3. 2017Understanding Black-box Predictions via Influence Functionscited by 1,741
  4. 2017Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environmentscited by 1,558
  5. 2020Sharpness-Aware Minimization for Efficiently Improving Generalizationcited by 1,417
  6. 2020Backdoor Learning: A Surveycited by 426
  7. 2017Adversarial Attacks on Neural Network Policiescited by 320
  8. 2022The Primacy Bias in Deep Reinforcement Learningcited by 247
  9. 2019Adversarial Policies: Attacking Deep Reinforcement Learningcited by 214
  10. 2024Deep Reinforcement Learning for Robotics: A Survey of Real-World Successescited by 176
  11. 2023The Dormant Neuron Phenomenon in Deep Reinforcement Learningcited by 174
  12. 2023Understanding plasticity in neural networkscited by 174
38 more
  1. 2019Adversarial Sensor Attack on LiDAR-based Perception in Autonomous Drivingcited by 152
  2. 2022Understanding and Preventing Capacity Loss in Reinforcement Learningcited by 125
  3. 2023Loss of Plasticity in Continual Deep Reinforcement Learningcited by 121
  4. 2020A Theoretical Analysis of Catastrophic Forgetting through the NTK Overlap Matrixcited by 103
  5. 2017Vulnerability of Deep Reinforcement Learning to Policy Induction Attackscited by 91
  6. 2024Disentangling the Causes of Plasticity Loss in Neural Networkscited by 86
  7. 2020On the Effectiveness of Mitigating Data Poisoning Attacks with Gradient Shapingcited by 83
  8. 2023Deep Reinforcement Learning with Plasticity Injectioncited by 74
  9. 2021BACKDOORL: Backdoor Attack against Competitive Reinforcement Learningcited by 68
  10. 2021Spectral Normalisation for Deep Reinforcement Learning: an Optimisation Perspectivecited by 65
  11. 2020Stealthy and Efficient Adversarial Attacks against Deep Reinforcement Learningcited by 63
  12. 2024Normalization and effective learning rates in reinforcement learningcited by 58
  13. 2021Adversarial Attacks On Multi-Agent Communicationcited by 49
  14. 2024Slow and Steady Wins the Race: Maintaining Plasticity with Hare and Tortoise Networkscited by 43
  15. 2024Best-of-Venom: Attacking RLHF by Injecting Poisoned Preference Datacited by 35
  16. 2022Not All Poisons are Created Equal: Robust Training against Data Poisoningcited by 30
  17. 2023BadRL: Sparse Targeted Backdoor Attack Against Reinforcement Learningcited by 30
  18. 2019How You Act Tells a Lot: Privacy-Leakage Attack on Deep Reinforcement Learningcited by 29
  19. 2022Adversarial Policies Beat Superhuman Go AIscited by 29
  20. 2024Weight Clipping for Deep Continual and Reinforcement Learningcited by 29
  21. 2023Revisiting Plasticity in Visual Reinforcement Learning: Data, Modules and Training Stagescited by 28
  22. 2024Is poisoning a real threat to LLM alignment? Maybe more so than you thinkcited by 28
  23. 2022BAFFLE: Hiding Backdoors in Offline Reinforcement Learning Datasetscited by 25
  24. 2020Stealing Deep Reinforcement Learning Models for Fun and Profitcited by 24
  25. 2024Plasticity Loss in Deep Reinforcement Learning: A Surveycited by 24
  26. 2019Design of intentional backdoors in sequential modelscited by 21
  27. 2023Reinforcement Unlearningcited by 17
  28. 2023Implicit Poisoning Attacks in Two-Agent Reinforcement Learning: Adversarial Policies for Training-Time Attackscited by 14
  29. 2024CLIBE: Detecting Dynamic Backdoors in Transformer-based NLP Modelscited by 13
  30. 2024Research on Autonomous Robots Navigation based on Reinforcement Learningcited by 12
  31. 2023ORL-AUDITOR: Dataset Auditing in Offline Deep Reinforcement Learningcited by 11
  32. 2022SoK: Rethinking Sensor Spoofing Attacks against Robotic Vehicles from a Systematic Viewcited by 9
  33. 2024SUB-PLAY: Adversarial Policies against Partially Observed Multi-Agent Reinforcement Learning Systemscited by 7
  34. 2024TrajDeleter: Enabling Trajectory Forgetting in Offline Reinforcement Learning Agentscited by 7
  35. 2025Temporal Logic-Based Multi-Vehicle Backdoor Attacks against Offline RL Agents in End-to-end Autonomous Drivingcited by 4
  36. 2024Slot: Provenance-Driven APT Detection through Graph Reinforcement Learningcited by 3
  37. 2026IntraGuard: Committee-Side Defenses Against Review Outsourcing to Commercial Chatbotscited by 2
  38. 2025Predictive Response Optimization: Using Reinforcement Learning to Fight Online Social Network Abusecited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Decision-Level Hijacking: Injecting Cognitive Bias into Large Language Models via Bit-Flip Attackscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026BehaviorGuard: Online Backdoor Defense for Deep Reinforcement Learning
  2. 2025Beyond Training-time Poisoning: Component-level and Post-training Backdoors in Deep Reinforcement Learning
  3. 2024Mitigating Deep Reinforcement Learning Backdoors in the Neural Activation Space
  4. 2026When Backdoors Meet Partial Observability: Attacking Real-World Reinforcement Learning
  5. 2025Backdoors in DRL: Four Environments Focusing on In-distribution Triggers
  6. 2022A Temporal-Pattern Backdoor Attack to Deep Reinforcement Learning
  7. 2025From Poisoned to Aware: Fostering Backdoor Self-Awareness in LLMs
  8. 2025Exposing Vulnerabilities in RL: A Novel Stealthy Backdoor Attack through Reward Poisoning ↗