cosedagarante, Rassegna Stampa ed Eventi, RassegnaStampa

Behind the agreements between publishers and AI is a huge privacy issue

#cosedagarante | From Le Monde to the Financial Times, from Axel Springer to the Associated Press, from the Wall Street Journal and the New York Post to the Times and the Sun, from the US to Europe, the list of publishers who have succumbed to OpenAI’s flattery, or more simply, to OpenAI’s money, by licensing their content to OpenAI is already long.

Pecunia non olet, they used to say, on the other hand the Latins, and business is business.

It is therefore good for many to sell their content to those who openly intend to use it – and are already doing so – to produce and have produced other content, even competing with that of the publishers who sell it to them.

Appropriate or not?

It is a business choice.

But also a legal one, since publishers cannot, of course, dispose of what does not belong to them.

This is where things get more complicated.

Publishers are licensing to OpenAI, allowing the commercial exploitation of the copyrights they assume to have on published content because they have acquired it from the authors.

And here is a first problem: are we sure that publishers all over the world have always bought from journalists and the plethora of different subjects (columnists, readers, legal advertisers, advertising investors for editorials) also the copyrights for a form of commercial exploitation – the training of algorithms – that simply did not exist until a few years ago?

It is a thorny issue, and one that both publishers and regulators have perhaps dealt with a little too superficially, as if it were all about money and business, and not also about rights and flesh-and-blood people.

But behind the agreements between publishers and algorithm factories there is another, even more complicated issue.

Newspaper articles and, more generally, the content of the digital editions of newspapers contain an infinite amount of personal data that can be traced back to billions of people: the protagonists of news, the subjects of legal advertising, the subjects of the millions of pieces of content from social networks that, for various reasons, end up on the pages of newspapers, the signatories of letters to the editor and the readers who seek advice from the experts who answer them on the pages of newspapers, and many more.

The publisher of a newspaper does not have any form of “property” of this personal data:

He, simply, can process them, generally without the consent of the data subject, solely in the exercise of his right to report or because he is asked to publish them by a public or private body that pays him for this purpose or for a series of other varied purposes.

And here is the second problem: in what capacity can a publisher sell this mountain of personal data to OpenAI or its competitors?

What do you think?

I started writing about this a few days ago in Agenda Digitale and I have returned to it today in Milano Finanza with some more thoughts.