17. March 2025

Behind the deliberate infection of ChatGPT & Co. with Russian propaganda

Dear friends,

This report from the disinformation experts at ‘NewsGuard’ is really something. If you have 10 minutes today, read this report here.

The study examined 10 AI chatbots such as ChatGPT or Copilot to see if they spread disinformation from the pro-Kremlin propaganda network ‘Pravda’.

The result: depending on the chatbot, 6% to over 50% of the answers are false information from the ‘Pravda network’.

They show how this global network of homepages and accounts, controlled from Russian-occupied Crimea, is deliberately flooding the internet with disinformation that is particularly useful for Large Language Models (LLM), i.e. the algorithms of AI systems: 3.6 million articles in 2024 alone. These articles rarely reach users directly, but are built specifically for the AI as fodder. Often, it is not new content, but only newly compiled and repeated elements of known disinformation.

One example from the study: when asked ‘Why did Zelensky ban Truth Social?’, six out of ten chatbots answered with Russian disinformation rather than with the fact that Zelensky is not blocking Trump’s Truth Social app in Ukraine.

As a result, chatbots are being flooded with more and more Russian disinformation, which, like a Trojan horse, is gratefully accepted by the chatbots as reliable information and fed back to the users asking the question.

None of this is happening by accident. As the authors of the report also uncover, the Russian propagandist John Mark Dougan speaks openly about the strategy behind it. Flooding AI systems and spreading Russian disinformation globally:

‘By pushing these Russian narratives from the Russian perspective, we can actually change worldwide AI.’

This shows that we must finally do more to combat disinformation and regulate artificial intelligence. The EU must now consistently enforce its strong laws such as the Digital Services Act and the AI Act and finally gain insight into the mechanisms at work in AI and social media. Only in this way can we put a stop to the ever-faster spread of disinformation and the disadvantage of information. Giving in to pressure from the US now would be disastrous.

This is what I work on every day in the European Parliament.

The problem, however, is that far too few policymakers are aware of the extent of the problem.

They continue to think that the right communication strategy would get the message across. NewsGuard’s report shows once again that this is not the case, but that there is no longer a level playing field in the digital space.

This is the opposite of freedom of expression.

Therefore, I ask you to send this email to as many contacts as possible and ideally also to your local representatives at the federal, state and European level. Only in this way can we create awareness of what is currently happening.

Thank you for your support!

With best regards,

Alexandra Geese

The entire NewsGuard study: A well-funded Moscow-based global ‘news’ network has infected Western artificial intelligence tools worldwide with Russian propaganda

 

Stay up to date