Two major international news agencies, Agence France Presse and Associated Press, have signed agreements with artificial intelligence players. Their conversational chats will now be able to use AFP and AP news reports in their responses. However, the question of the moral rights of the authors of the dispatches used by these generative AIs is not raised.
The first agreement is that of AFP with Mistral, the French champion ofartificial intelligencewhich will be able to have access to the agency’s 38 million dispatches since 1983. On paper, it is an agreement which only presents advantages. Mistral AI, founded two years ago, benefits from the expertise of one of the three major global press agencies, in six languages. Its solution, Chat, will be able to support its responses with verified, sourced content, which constitutes a competitive advantage while other major European media, such as Axel Springer, El Pais Or The World have already signed agreements with ChatGPT. Mistral will provide a dynamic source of income for the AFP, which is thus keeping pace with AI.
Mistral, a start-up that looks a lot like its American cousins
Mistral has renounced a logic ofopen source to enter into a private agreement last year with Microsoft, which acquired a stake in its capital and which is also a 49% shareholder in Open AI. Above all, Mistral put a lot of weight on the French government during negotiations on the AI Act last year so that European innovation was not hampered by copyright. But have the journalists, the editors, given their agreement for the dispatches to be used to feed the responses of a technological solution? Is there not a risk of seeing the tool replace the media themselves on these platforms?
Also readThe gray areas of the European AI Act
The Associated Press has a deal with Google’s Gemini
This agreement clearly shows that the credibility of generative AI through reliable news content is crucial for these platforms. Apple has also just uninstalled its AI news summary tool after errors made from content from the BBC which complained about it.
AP, the Associated Press, already had an agreement with Open AI, on ChatGPT, since 2023. In the United States, the New York Timesalong with other publishers, sued this AI actor for copyright infringement. He estimates the damage at several billion dollars. This is also the logic of professional press organizations in France who want to obtain a collective agreement.
The risk may be, according to some, the very existence of the press and, for an agency like AFP, to depend on these new revenues which, as shown by Meta’s decision to renounce thefact-checking, can stop overnight.
Also readUnited States: the end of fact-checking on Meta, a challenge for the media