With the recent rise of artificial intelligence, technology companies such as Google and Microsoft, as well as others specialized in the sector such as OpenAI, have found themselves in a situation where they had to find an infinite amount of data sources. train the algorithms on which the services they offer are based. And when I talk about infinity, I really mean volumes that are several orders of magnitude beyond those that have been worked with for virtually any other purpose up until now.

This can be a problem because it has forced these companies look for as many data sources as possible, something in which the Internet played a key role. From niche sites to forums, from media to social networks, until now all content published on the Internet has been susceptible to being swallowed by one or more algorithms during their training. Of course, with the risk that a posteriori, if we are talking about generative models, they could reproduce it in a practically identical way in which they received it during their “formation”.

How do you know, This has led to a number of movements proposing that limits be set in this regard., among other things, because some of the material used is covered by certain copyrights and therefore cannot of course be used commercially or reproduced. And then we find slightly more special cases like Twitter, which was quite angry and even wanted to condemn OpenAI because, according to Elon Musk, it used huge sets of tweets to train its language models in those based on both GPT in their iterations and ChatGPT .

Aware of this problem, but also of the need to feed the training processes, Google has made it clear that it will use personal publications to train its models. Specifically, it did so by making some changes to the English version of the privacy terms that previously indicated that personal contributions could be used to train language models, citing its translator as an example, and now specifically mentioning “AI models” and mentioning services like Bard and Cloud AI .

At this point it is important, yes, to be clear Google is talking about personal posts, not personal data. In other words, all of our private content will continue to maintain this state and therefore will not be used for these purposes. And given that the terms have only been updated in their English version, that is, the one that is primarily aimed at the company’s internal market (the United States of America), it makes sense to think that, at least for now, this change will not apply internationally and especially in Europe, where the search engine company is working to launch its AI-powered services by adapting to a complex legal framework that has already begun to take shape.