Trading Devil RL: Backdoor attack via Stock market, Bayesian Optimization and Reinforcement Learning
Orson Mengara
TL;DR
The work addresses the security risks of backdoor data-poisoning in FinanceLLM-RL systems by proposing FinanceLLMsBackRL, a backdoor framework that uses diffusion-based Bayesian optimization and Navier–Stokes-inspired dynamics to craft stealthy triggers. It couples this attack design with a dynamic-systems–driven detection approach, including Lyapunov-function-based stability analysis and bootstrap statistics, to identify poisoned models. The paper contributes a novel attack model, extensive evaluation on audio-transformer victims, and a universal defense mechanism that leverages dynamical-systems theory and meta-learning to enhance robustness. These results highlight tangible security vulnerabilities in LLM-RL financial pipelines and provide a principled pathway toward monitoring and mitigating backdoor risks in deployed AI systems.
Abstract
With the rapid development of generative artificial intelligence, particularly large language models a number of sub-fields of deep learning have made significant progress and are now very useful in everyday applications. For example,financial institutions simulate a wide range of scenarios for various models created by their research teams using reinforcement learning, both before production and after regular operations. In this work, we propose a backdoor attack that focuses solely on data poisoning and a method of detection by dynamic systems and statistical analysis of the distribution of data. This particular backdoor attack is classified as an attack without prior consideration or trigger, and we name it FinanceLLMsBackRL. Our aim is to examine the potential effects of large language models that use reinforcement learning systems for text production or speech recognition, finance, physics, or the ecosystem of contemporary artificial intelligence models.