Fake News, Real Crashes: Information Injection Attacks on LLM Financial Trading Agents
We present a systematic attack benchmark for information injection against LLM trading agents.
ProofPaper ↗
Key points
- Large language model (LLM) agents now trade real capital using news as evidence, yet they cannot reliably tell fabricated headlines from real news.
- Because they read the same feeds under near-identical policies, a single fabricated post can push them into the same trade at the same moment.
- Unlike prompt injection, which targets the model's instructions, information injection targets the evidence the agent trusts.
- The benchmark spans five injection vectors and seven event categories, evaluated over 10,000 decisions on 15 U.S. stocks.
Sources (1)
- [1]Fake News, Real Crashes: Information Injection Attacks on LLM Financial Trading AgentsarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 19, 01:58 PM
We present a systematic attack benchmark for information injection against LLM trading agents.
Large language model (LLM) agents now trade real capital using news as evidence, yet they cannot reliably tell fabricated headlines from real news.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 19, 2026Productive State-Space Exploration for Multi-turn RL in LLM Agents
- Oct 19, 2026AdaExploreBench: Benchmarking LLM Agents for Proactive Exploration and Test-time Adaptation in Evolving Environments
- Oct 18, 2026SoK: PIAtlas: Decomposing the Unbounded Prompt Injection Attack Space
- Oct 17, 2026AgentSoV: Virtual Sensing for LLM Multi-Agent Systems via Stream of Variation
- Oct 17, 2026Pair-Aware Optimization: Coordinating Exploration and Refinement for Jailbreaking LLM Agents
- Oct 16, 2026SPIDES: Information-Flow Control for Browser Agents