We saw it first with GitHub Copilot and then with Stable Diffusion . Training generative artificial intelligences with raw data and without paying for licenses is beginning to be the reason for lawsuits by content (and code) creators, and now that same problem is beginning to surface in conversational AI engines such as ChatGPT.

ChatGPT feeds on media such as the Wall Street Journal or CNN. Now those media want to get paid for it.
ChatGPT feeds on media such as the Wall Street Journal or CNN. Now those media want to get paid for it.

What has happened Jason Conti, general counsel of the Dow Jones division at News Corp, made it clear: “Anyone who wants to use the work of Wall Street Journal journalists to train artificial intelligence should properly license the rights to do so.” Or what is the same: if ChatGPT or (Bing with ChatGPT, or the hypothetical Bard) want to use their contents to be trained and polished, they will have to license them first.

Veiled threat . This manager went even further and as indicated in Bloomberg added that “we take the misuse of the work of our journalists very seriously, and we are reviewing this situation.” A journalist named Francesco Marconi precisely asked ChatGPT for its sources and among them were both the WSJ and CNN. The first case is even more striking, especially since WSJ applies a very rigorous paywall.

CNN joins the WSJ . Sources close to CNN revealed in that same Bloomberg article that this media outlet also believes that ChatGPT is violating the terms of service by being trained with data collected from its articles.

OpenAI, you should pay . That is the conclusion both news groups seem to reach. CNN, owned by Warner Bros. Discovery Inc., intends to negotiate with OpenAI for a license to use that content in the conversational engine.

A debate that comes from afar . The media already faced a similar situation years ago with the famous ’ Google Tax '. The introduction of the Intellectual Property Law meant that Google News and other link aggregators had to pay the authors of the content they linked to. In October 2014, Canon AEDE arrived , and with it, the closure of Google News in Spain, which ended up being active again in June 2022 after the approval of the ’ Iceta Law '.

Looking for a fair model . Search engines like Google’s were criticized for taking advantage of content generated by media and creators without giving direct income for it. Media companies and creators, yes, obtained traffic to their media, although this has also been criticized for the way in which, for example, Google almost makes it almost unnecessary to leave the search engine .

The creators before a world of gluttonous AIs . Generative AIs don’t send traffic or indirect revenue to creators at this time—they don’t in Copilot or Stable Diffusion/DALL-E 2/Midjourney—and now the gigantic search and advertising business may be compromised as well. It therefore seems necessary to have a commitment so that the AIs are useful, but giving something in return to those who have provided the content with which they are trained and improved.