For Ezra Eeman, head of AI in media at the World Association of News Media Publishers (WAN-IFRA), it is in the media’s best interest to focus on improving direct access to their content with the help of AI.
JDN. Large LLMs only sign partnership agreements with major media: how can others survive in the face of chatbots?
Ezra Eeman. The AI giants are not interested in entering into partnership agreements with all publishers, they do not need them. AI represents a significant threat to media whose business model remains dependent on online traffic. The latter must therefore rethink their model. The era of earning money from massive traffic is over. The development of news media will now come primarily from exclusive content and services and a direct relationship with their readers.
The AI giants, however, do not seem embarrassed to steal publishers’ content without their authorization…
Publishers are indeed able to notice that these robots come to them and take content. Unless they offer truly exclusive content, it is however very difficult for them to demonstrate that these chatbots transform this content into output, given that these companies do not share their data. This is why it seems fundamental to me that publishers protect their exclusive content from crawlers, it is their main negotiating lever with these platforms.
Interesting avenues are emerging so that a larger number of publishers can participate in this monetization economy with LLMs, without having to sign one-to-one partnership agreements. I am thinking in particular of certain new standards, including the RSL (for Really Simple Licensing, editor’s note). The RSL consortium offers a sort of digital signature which, based on the site code, tells the crawler whether it can use the site’s content and under what conditions. This goes much further than robots.txt, which simply indicates that access is not authorized. Cloudflare offers a protocol similar to the RSL standard. Furthermore, we can even imagine that tomorrow, if an advertisement were to be placed next to content taken from a publisher site, then the latter should have the right to receive a percentage of this advertising revenue. But for this to work, AI companies must adopt these standards, which is not yet the case. However, I remain hopeful as pressure for their adoption is mounting, at least here in Europe.
What importance could this monetization with AI take on in the media economy in your opinion?
We should not be under any illusion, the monetization by publishers of their content with AI will generate small amounts, the latter will enrich the diversification portfolio but will remain very secondary. No media will be able to base its economic model on this type of income. Even if the adoption of an RSL type protocol became widespread tomorrow, it would never take on the same importance for a publisher as its revenues from subscriptions, direct traffic and even advertising.
What AI-based innovations can support the development of media, and the press in particular, in these difficult times?
With AI and reader behavioral data, publishers can offer a much more personalized experience taking into account their preferences (time dedicated to content, time of day, preferred topics, etc.). Publishers will also be able to increasingly make available, for the same content, different formats: audio, video, summaries, other articles linked with much more relevance. Tomorrow they could even offer personal agents to their readers who would be responsible for preparing the day’s edition or newsletter for them, taking into account their preferences and the content available from the media! It is now a matter of inventing more clever and creative ways of bringing content to readers and retaining them.
You are also director of strategy and innovation at NPO, the Dutch public broadcaster. How do these same challenges threaten your sector?
The situation among broadcasters is very different from print media. The latter are in a more fragile position because it is difficult to prove that an article was pumped by an AI, unless it is truly unique, whereas it is much more complicated to do so with video or audio. The challenge for our sector, however, is that with AI many more people can produce very good quality videos with disconcerting ease.
What opportunities does AI offer to broadcasters like you?
At NPO (followed by around 87% of the Dutch population, editor’s note)we produce all types of content: news, entertainment and fiction. AI helps us adapt our content much more easily to different idioms (English being used almost more in big cities than Dutch in our country) and formats, including on social, a channel that we explore a lot and that AI helps us moderate.
