The measure was approved at the plenary session of the EP in Strasbourg, with 423 votes in favour, 57 against and 174 abstentions, and aims to amend certain rules currently in force under European AI law.

The EP said that, if it receives final "green light" from the EU Council, the law will prohibit "AI systems that generate child sexual abuse material or that create images, videos and audios depicting intimate parts of an identifiable person, or sexually explicit activities, without their consent," according to a statement.

"Suppliers will not be allowed to place these systems on the EU market unless they have adequate technical safeguards to prevent the creation of this material. The ban also applies to users who resort to these systems for these purposes," the EP indicated.

The institution added that companies now have until 2 December to update their systems and ensure they comply with these new rules on sexual images.

This proposal had already been agreed upon by the EP and the EU Council, which represents the governments of the 27 member states.

However, to enter into force, the measure will have to be formally adopted by the EU Council.

The proposal came after the European Commission decided in January to open an investigation into Grok, an AI tool from the social network X, for disseminating sexually explicit images, including content that could constitute child sexual abuse.

According to a report by the Centre for Combating Digital Hate (CCDH) and the American newspaper The New York Times, in January, Grok allegedly flooded that network with approximately three million sexualised images over 11 days, including 23,000 of children and 1.8 million of women.

In addition to the ban on AI systems that generate sexually explicit images, this change in European law also includes measures to simplify the regulatory framework currently applied to AI.

Thus, the proposal now provides for the possibility, in both general-purpose and high-risk AI systems, "of processing personal data whenever strictly necessary to detect and correct biases, with adequate safeguards."

It also eliminates the "overlapping of AI requirements applicable to machines, components and accessories," with the AI ​​law clarifying that "they only need to comply with sectoral legislation, while ensuring an equivalent level of health and safety."