Israel-Linked Influence Campaign Manipulated AI Algorithms

A state-funded influence operation sought to shape public perception regarding the Israeli-Palestinian conflict by manipulating artificial intelligence systems and search engine results. An investigation into the Hanover Institute for Public Policy uncovered the campaign, which employed advanced digital communication strategies to influence narratives within AI models.

Documents filed under the Foreign Agents Registration Act (FARA) link this operation to LaPam, Israel’s state advertising agency. LaPam commissioned the New York-based firm Piro Inc. for a digital communications strategy valued at $900,000. This initiative represents an evolving approach to narrative influence, moving beyond traditional media engagement to directly manipulate information ecosystems.

Content Generated for AI Platforms

Hanover Institute

The campaign utilized a method termed 'AI Story Optimization,' generating over 560,000 words across 124 academic-style reports within nine days. This content was specifically designed to mimic neutral research, aiming for its perspective on sensitive subjects, including military conduct and humanitarian conditions in Gaza, to be prioritized by large language models and search algorithms.

The strategy exploited the reliance of AI systems on data that appears authoritative. By presenting advocacy as independent expertise, the operation sought to embed its framing of the conflict into the core responses of artificial intelligence platforms, influencing public perception through automated information synthesis.

AI System Vulnerabilities

Major AI platforms

Independent third-party outlets conducted tests confirming the effectiveness of this method. Major AI platforms, including ChatGPT and Perplexity, were found to cite this fabricated material when responding to user queries. This validation underscored the campaign's success in infiltrating the information sources utilized by these advanced AI models.

The incident highlights systemic vulnerabilities in how AI models verify sources and authenticate information. It raises significant concerns about the potential for foreign entities to influence public discourse by manufacturing and disseminating information designed for absorption and reproduction by AI systems, thereby shaping perceptions on critical global issues.