LLM Summaries Can Distort Financial Decisions, Study Finds

Hoyoung Lee, Suhwan Park, Seunghan Lee, Jun Seo, Jaehoon Lee, Sungdong Yoo, Minjae Kim, CheolWon Na, Zhangyang Wang, Zach Golkhou, Minkyu Kim, Sotirios Sabanis, Alejandro Lopez-Lira, Dhagash Mehta, Soonyoung Lee, Chanyeol Choi, Wonbin Ahn, Yongjae Lee· June 30, 2026 View original

Summary

Research shows that large language model (LLM) compression of financial documents can alter investment judgments, even when summaries appear factual. The study introduces "information fidelity" to evaluate how well compressed content preserves decision-relevant context.

Financial professionals often rely on LLMs to summarize vast amounts of information, but new research highlights a critical risk: these summaries can inadvertently distort the underlying investment decisions. The study defines this issue as "information fidelity loss," where LLM compression, despite producing fluent and factually plausible text, changes the judgment an investor would make based on the original source. This fidelity loss often stems from "decontextualization," where key evidence is retained but separated from crucial caveats, and "model dependency," where different LLMs yield varying interpretations. To counter this, the researchers propose Agentic Context Compression, a method that generates multiple summary candidates and audits their discrepancies against the original document to ensure decision-relevant context is preserved.

Why it matters

Professionals using LLMs for financial analysis must be aware that even accurate-looking summaries can lead to incorrect decisions, necessitating robust validation methods.

How to implement this in your domain

  1. 1Implement a multi-LLM summarization strategy for critical financial documents to compare outputs.
  2. 2Develop internal auditing processes to cross-reference LLM-generated summaries with original source material for key decision points.
  3. 3Train financial analysts to identify common patterns of fidelity loss, such as decontextualization, in LLM outputs.
  4. 4Integrate tools that highlight discrepancies between different LLM summaries or between summaries and original text.

Who benefits

Financial ServicesInvestment BankingAsset ManagementLegalConsulting

Key takeaways

  • LLM-generated financial summaries can subtly alter investment decisions despite appearing accurate.
  • Information fidelity loss occurs through decontextualization and model dependency.
  • Evaluating LLM compression should prioritize preserving decision-relevant context, not just factual accuracy.
  • Agentic Context Compression offers a method to audit and improve summary reliability.

Original post by Hoyoung Lee, Suhwan Park, Seunghan Lee, Jun Seo, Jaehoon Lee, Sungdong Yoo, Minjae Kim, CheolWon Na, Zhangyang Wang, Zach Golkhou, Minkyu Kim, Sotirios Sabanis, Alejandro Lopez-Lira, Dhagash Mehta, Soonyoung Lee, Chanyeol Choi, Wonbin Ahn, Yongjae Lee

"arXiv:2606.29251v1 Announce Type: new Abstract: Financial decision-makers face more information than they can directly inspect, making context compression necessary. Yet when large language models (LLMs) compress financial source material, they can alter the investment judgment s…"

View on X

Originally posted by Hoyoung Lee, Suhwan Park, Seunghan Lee, Jun Seo, Jaehoon Lee, Sungdong Yoo, Minjae Kim, CheolWon Na, Zhangyang Wang, Zach Golkhou, Minkyu Kim, Sotirios Sabanis, Alejandro Lopez-Lira, Dhagash Mehta, Soonyoung Lee, Chanyeol Choi, Wonbin Ahn, Yongjae Lee on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses