A Nation State Just Built a Fake Think Tank to Manipulate AI Chatbots — and It Changes Everything

A recently exposed investigation by Responsible Statecraft revealed that Israel has created a fake think tank, the Institute for Research on Middle East Policy, designed to manipulate AI chatbots into spreading propaganda. This development marks a new frontier in information warfare, known as AI model poisoning, a state-level disinformation strategy.

The operation published policy papers and reports with a pro-Israel narrative, aiming to influence AI-generated information. Traditional disinformation targets humans, while AI-targeted manipulation focuses on AI systems that humans increasingly rely on for information. The attack follows a clear pipeline: creating a front think tank, producing content, seeding the ecosystem, and then AI ingestion and amplification.

The cost of creating such a think tank is minimal, allowing state actors or corporations to potentially create multiple fronts. While AI labs are aware of this threat and are developing defensive measures, these are reactive and may not be effective against sophisticated operations. Developers should implement verification systems, cite sources, and monitor for coordinated campaigns to avoid being misled by AI-targeted disinformation.

This new threat demonstrates that AI chatbots, which are becoming primary information layers for many people, could be manipulated silently, actively misleading users.