Artificial intelligence is one of the most talked about technologies. In Europe, you can now access a large number of platforms that allow you to do all kinds of things. From creating melodies with Suno.IO to writing emails with Gemini or synthesizing texts with ChatGPT.
Of course, the possibilities they offer are many, and one of them is to create images, either from another image or a description, or using only that description proposed by the user. However, This particular action is causing quite a bit of controversy..
And there are times when this functionality is used for negative purposes, something that the platforms themselves already model so that it does not happen, although there are exceptions such as Elon Musk’s new AI, which is quite lax about these limits. But what are the limits of creating images through artificial intelligence?
Copyright
Since the expansion of artificial intelligence models, many alternatives have emerged whose function is to create images using this technology. This can have a a large number of positive aspects,
However, some of them had to put restrictions in place along the way, as The possibilities it offers require regulation so that it is used correctly.. Copyright is one of the limits of the possibilities of AI, not only in terms of its creations, but also in terms of its training. In this sense, European regulations specify that “any use of content protected by copyright requires the authorization of the holder of the rights in question.”
The regulation still has a long way to go, but it underlines the responsibility of AI model providers in terms of training. On the other hand, as far as creation is concerned, if the image generated with this technology represents a substantial transformation of an original work, for example, it could be considered a derivative work and, in some cases, even be authorized.
In many cases, these types of tools already have some limitations, such that they can only generate licensed images, so that the risk of violating the copyright protection of a specific work is reduced as much as possible. If AI generates a new image very similar to an existing work
Nudes using AI
Some AI models can not only create images from a description, but also modify one or more images using others, which some people have used to create what is called a deepfake. This is a modification of a person’s image to make them look like another, and this was used to artificially remove other people from photographssomething totally illegal in Europe.
Some artificial intelligence platforms, such as Microsoft’s Designer, have chosen to establish limits on user queries so that they cannot use it for such purposes. Despite this, episodes have already taken place in Europe, such as the Almendralejo case, in which AI was used to carry out these montages with minors and which have put the debate back on the table.
There are other more recent cases with the same problem, and in this sense, the National Police confirms that the use, possession and dissemination of these altered images can be “constitutive of a crime.” In Europe, this act can be framed in several crimes included in the Penal Code.However, regulation at European level from which different nations could legislate would also be positive.
Elon Musk’s artificial intelligence model, Grok-2, integrated into X, is rightly harshly criticized for its few ethical limits. And, if there are demands that other AI models would refuse, this one accepts them, highlighting the need to regulate and limit this type of technology.
Microsoft, one of the world’s largest technology companies, has asked the US government in a letter to regulate such images in order to “protect the public from abusive content generated by AI,” which is precisely the title of the letter.
And this technology is capable of doing good deeds, but also doing harmful and damaging things to the image of others. And not only that, but it has been used several times to commit scams, impersonating other people in the famous CEO scam.