French newspapers and magazines are launching an offensive against public online databases used to train generative artificial intelligence applications that they say use content without compensating the authors, putting the France's professional journalism sector's economic model at risk.
Two professional organisations representing 800 newspapers and magazines employing over half of journalists in France announced Monday that they are taking “coordinated action” against public datasets used to train generative artificial intelligence services, such as ChatGPT.
Public-access datasets
The Apig, the general news medial alliance, and the Sepm, the magazine publisher's union, aim to remove their members’ content from Common Crawl, C4 and Oscar – public-access datasets created by bots that “crawl” the internet.
The groups denounce what they say is a system that collects and distributes copyrighted articles “without authorisation or any access restrictions", and allows generative AI service providers "to source press material through these intermediaries, avoiding any direct negotiations with publishers and respect of intellectual property rights".
Their strategy involves identifying the presence of copyrighted content in the datasets, formally requesting their removal, and preparing legal action against those who have profited from their use.
French court blocks Google project to limit news content in searches
The legal framework to protect copyrighted material is not solid, Apig CEO Pierre Petillault told RFI, and the political will to support authors rights clashes with France’s interest in technological innovation.
"There is this tension between innovation and intellectual property that unfortunately sometimes leads public authorities to be a little complacent towards large digital platforms,” he said.
“There is the temptation to promote innovation” and allow European AI companies the leeway to compete with the United States.
The impact of new technologies
However, the organisations argue that protecting the "professional information" sector is crucial.
Journalism and "professional information" requires investment and content must be compensated, the groups said.
Their initiative aims for "a fair sharing of the value generated by these new technologies.
It continued: "In a context where professional journalism's economic viability is already fragile, this unauthorised capturing of value represents a direct threat to the quality of information, and ultimately, to democracy".
EU begins rollout of new AI rules with tech giants split on compliance
In February, five press organisations – including the Apig and the Spem - representing more than 3,000 newspapers and magazines called on the public authorities to “impose a dialogue” between AI companies and the media, to “put an end to the plundering” of their content and preserve France’s information ecosystem.
