Overseas malign information operations by foreign actors seek to undermine U.S. strategic interests. Such operations often involve using classical propaganda techniques to shape narratives, shift public discourse, and undermine other nations' national security. This report describes the authors' efforts to use large language models to detect the use of classical propaganda techniques in hundreds of articles published in overseas media.

Defending American Interests Abroad

Early Detection of Foreign Malign Information Operations

Christopher A. Mouton, Caleb Lucas, Shaun Ee

ResearchPublished Apr 2, 2025

Overseas malign information operations by foreign actors seek to undermine the strategic interests of the United States. These operations are intended to manipulate the global information environment for geostrategic purposes by disseminating false or misleading information to shape narratives, shift public discourse, and undermine other nations' national security. These operations often leverage classical propaganda techniques, such as name-calling, glittering generalities, transfer, testimonial, plain-folk, card-stacking, and bandwagon.

In this report, the authors explore methods to detect underlying instances of propaganda devices that might be indicative of broader malign information operations abroad by analyzing hundreds of articles from multiple overseas media outlets. Specifically, the authors explore using large language models (LLMs), which present an opportunity for the detection of overseas propaganda, disinformation, and misinformation targeting foreign audiences. These instances can be indicative of broader attempts by foreign adversaries to undermine U.S. credibility, influence, and alliances abroad.

Unlike traditional methods that rely heavily on specific linguistic indicators or network analysis, LLMs have increasingly large context windows, which offer the ability to incorporate the broader context of the text. The authors assess the performance of these detection tools to gain insights into the capabilities of these technologies to identify overseas propaganda. This represents a first step in preventing malign actors from exploiting the overseas information environment to pursue their strategic goals at the expense of other nations.

Key Findings

  • LLMs, which can interpret the broader context of text, can effectively automate overseas propaganda detection through a new implementation paradigm. This detection can be a useful first step is characterizing malign information operations that seek to undermine the strategic interests of the United States abroad.
  • When prompted to label text according to a set of classical propaganda techniques, LLMs can provide indicators (i.e., yes or no) of whether the text employs each technique, thereby partially mitigating subjectivity.
  • A fine-tuned model, trained on foreign propaganda manually coded for the presence of classical propaganda techniques, was shown to have effective performance in detecting the use of propaganda devices. This suggests that LLMs are a potentially useful detection method of foreign adversaries attempts to exploit other nations for their own objectives.

Recommendations

  • Future research could develop a corpus of identified propaganda instances to illuminate the aims of individual and collective overseas information campaigns conducted by hostile actors, thereby informing strategies to counteract malign information operations abroad that seek to undermine the national security interests of the United States.
  • Future research could include more non-English sources because much of the harmful information activity targeting foreign audiences is conducted in languages other than English, as malign actors seek to exploit other nations for their own strategic objectives.

Topics

Document Details

Citation

RAND Style Manual

Mouton, Christopher A., Caleb Lucas, and Shaun Ee, Defending American Interests Abroad: Early Detection of Foreign Malign Information Operations, RAND Corporation, RR-A2853-1, 2025. As of April 8, 2025: https://www.rand.org/pubs/research_reports/RRA2853-1.html

Chicago Manual of Style

Mouton, Christopher A., Caleb Lucas, and Shaun Ee, Defending American Interests Abroad: Early Detection of Foreign Malign Information Operations. Santa Monica, CA: RAND Corporation, 2025. https://www.rand.org/pubs/research_reports/RRA2853-1.html.
BibTeX RIS

Research conducted by

This research was conducted within the Acquisition and Technology Policy Program of the RAND National Security Research Division (NSRD) which operates the RAND National Defense Research Institute (NDRI). This research was made possible by NDRI exploratory research funding that was provided through the federally funded research and development center (FFRDC) contract and approved by NDRI's primary sponsor.

This publication is part of the RAND research report series. Research reports present research findings and objective analysis that address the challenges facing the public and private sectors. All RAND research reports undergo rigorous peer review to ensure high standards for research quality and objectivity.

This document and trademark(s) contained herein are protected by law. This representation of RAND intellectual property is provided for noncommercial use only. Unauthorized posting of this publication online is prohibited; linking directly to this product page is encouraged. Permission is required from RAND to reproduce, or reuse in another form, any of its research documents for commercial purposes. For information on reprint and reuse permissions, please visit www.rand.org/pubs/permissions.

RAND is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.