European media consortium launches Euro chatbot to counter fake news

ChatEurope will feature only verifiable sources

A group of major European media organisations has launched ChatEurope - a chatbot aimed at combating online disinformation.

A consortium of media brands led by Agence France-Presse and including news organisations including France Médias Monde, Deutsche Welle, El País, and Romania’s RFI, have launched ChatEurope.

The difference between ChatEurope and other consumer-facing chatbots like ChatGPT, is that ChatEurope is fed only verifiable and trusted sources. Users asking it a question can do so in the knowledge that the answer will be factually correct and free from bias – at least, that’s the theory.

The project is also backed by the European Commission.

The chatbot was designed by DRUID AI, a Romanian enterprise AI platform, and uses a LLM developed by French AI company Mistral. It’s hosted on infrastructure built by Paris-based open-source software provider XWiki.

Andreea Pleșea, co-founder and CEO of DRUID AI, said the tool enables “natural, human-like conversations” that help people access credible information in their own language. Languages covered at present include English, French, German, Italian, Polish, Romanian, and Spanish.

The LLM behind ChatEurope is trained on news articles produced by respected news outlets from the 15 consortium members. The chatbot therefore provides answers to news and geopolitics questions with full source citations.

It is a clear and welcome attempt to counter the tide of unverified and fake news, amplified and disseminated on platforms such as X, TikTok, and the biggest of them all – Facebook.

Concern about online misinformation and disinformation has grown in line with the proportion of individuals using social media as a primary – or sole – source of news. At the same time, social media platforms have scaled down fact checking operations (often for nakedly political reasons) and transferred the responsibility to users via community notes.

At the same time, trust in mainstream media has collapsed and a continual stream of conspiracy theories, hate speech and lies are pumped out via social media and the chatbots which can be easily manipulated to lie to users.

Harnessed by hostile states and the populist right, the consequences of misinformation poisoning the media ecosystem have already been profound. This is the first time such concern has translated into action.

“Media urgently need to adapt in the face of pervasive disinformation and the advent of artificial intelligence,” said Christine Buhagiar, AFP’s director of development and diversification. “ChatEurope will be a game-changer when it comes to information on European affairs.”

It’s not difficult to predict how bad faith actors will try to discredit ChatEurope. The endorsement of the EU might prove to be a millstone if opponents can portray it as riddled with pro-European algorithmic biases and as a tool to censor dissenting points of view.

The creators of ChatEurope claim that the bot is free of such bias. Given the stakes, a high degree of model transparency would be a sensible way to prove that in the face of the inevitable counterpunch.