# AI chatbots and foreign disinformation management

> Live situation record from CLSTR: https://clstr.news/situations/ai-chatbots-and-foreign-disinformation-management
> Updated: 2026-09-02T06:44:40.000Z. Sources: 22. Developments: 2.

Research indicates that AI chatbots are showing varying levels of effectiveness in managing foreign propaganda and disinformation. A study conducted between December 2025 and July 2026 found that models successfully identified and refuted misinformation from China, Iran, and Russia approximately 75% of the time. For example, Google’s Gemini correctly identified false Russian claims regarding the bombing of a Ukrainian monastery as part of a disinformation campaign.

However, subsequent investigations by Reporters Without Borders (RSF) have highlighted vulnerabilities in how these tools handle sanctioned content. Tests conducted in mid-2026 revealed that several major chatbots, including OpenAI’s ChatGPT, Anthropic’s Claude, and X’s Grok, were able to bypass European Union sanctions by providing access to content from outlets like Sputnik and RT. While Google’s Gemini and Mistral’s Vibe eventually provided access to sanctioned content after repeated requests, Meta AI was noted as the only chatbot that complied with sanctions by refusing the prohibited information.

In updated findings released in September 2026, RSF reported that tests in June and August showed ChatGPT, Claude, and Grok could reproduce content and provide complete press reviews from sanctioned Russian outlets without consistently noting their status. Consequently, RSF has called on the European Commission to investigate OpenAI regarding systemic disinformation risks and has suggested infringement procedures for other providers. OpenAI stated it is reviewing the findings, and Anthropic is conducting its own assessment regarding Claude.

## Claims

- RSF investigation found that several AI chatbots allow access to sanctioned Russian state media. (corroborated by 3 sources)
- Meta AI was the only chatbot tested that refused to provide access to the sanctioned media. (corroborated by 3 sources)
- OpenAI is reviewing the findings provided by RSF. (corroborated by 3 sources)
- Google Gemini and Mistral's Vibe eventually provided access to the sanctioned content after initial resistance. (corroborated by 2 sources)
- RSF called on the European Commission to investigate OpenAI regarding systemic disinformation risks. (corroborated by 2 sources)
- The AI chatbots provided complete press reviews without mentioning sanctions. (corroborated by 2 sources)
- Anthropic is conducting an assessment of situations related to Claude. (single source)

## Timeline

### 2026-09-02: Reporters Without Borders finds AI chatbots bypass EU sanctions on Russian media

Reporters Without Borders reports that AI chatbots like ChatGPT, Claude, and Grok bypass EU sanctions by providing access to Russian state media, while Meta AI remains compliant.

11 sources. https://clstr.news/cluster/reporters-without-borders-finds-ai-chatbots-bypass-eu-sanctions

### 2026-08-31: AI chatbots show ability to identify foreign propaganda

A study shows AI chatbots can successfully identify and refute foreign propaganda from Russia, China, and Iran approximately 75% of the time, according to research by NewsGuard and NPR.

11 sources. https://clstr.news/cluster/vietnam-airlines-expands-kansaihanoi-flight-service

---
Cite as: AI chatbots and foreign disinformation management. CLSTR, https://clstr.news/situations/ai-chatbots-and-foreign-disinformation-management