Newest papers

Fake News, Real Crashes: Information Injection Attacks on LLM Financial Trading Agents

Mengxiao Wang, Nitesh Saxena · 2026 · arXiv

DOI: 10.48550/arXiv.2610.26438 · open-access PDF

Portuguese

Os agentes do modelo de idioma grande (LLM) agora negociam capital real usando notícias como evidência, mas eles não conseguem dizer com confiança os títulos fabricados de notícias reais. Porque eles lêem os mesmos feeds sob políticas quase idênticas, um único post fabricado pode empurrá-los para o mesmo comércio no mesmo momento. Nós apresentamos um benchmark de ataque sistemático para injeção de informação contra os agentes de negociação LLM. Ao contrário da injeção rápida, que visa as instruções do modelo, a injeção de informação visa a evidência que o agente confia. O benchmark abrange cinco vetores de injeção e sete categorias de eventos, avaliadas em ~1

Machine translation (Meta M2M100); the English original follows.

Abstract

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. We present a systematic attack benchmark for information injection against LLM trading agents. 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 ~1

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Is this paper real? This page was built from OpenAlex and arXiv records, not generated. Software can verify any reference in one call: GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.26438, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txt

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange