EU member states and the European Parliament agreed overnight to ban artificial intelligence services in the bloc that allow people to be “undressed” without their consent.
The initiative follows, in particular, the introduction a few months ago of a feature on Grok, the AI assistant developed by Elon Musk’s xAI, which allows users to ask it to create fake nude images -deepfakes- of adults and children based on real images, without the consent of those depicted.
The issue caused an outcry in several countries and led to the launch of an investigation in the EU.
On Tuesday, Italian Prime Minister Giorgia Meloni condemned fake AI-generated images of herself, describing such deepfakes as a “dangerous tool”.
According to the European Parliament, the new EU ban covers systems that allow the creation of images, videos and audio of a child sexual abuse nature, or material that depicts the intimate parts of an identifiable person or shows them engaging in sexual activities without their consent.
The new regulation will apply from 2 December 2026. From that date, AI systems will therefore have to be equipped with safeguards preventing them from generating such content.
The measure was adopted as part of a review of the EU’s artificial intelligence legislation, the AI Act, a landmark law formally adopted two years ago.
On the same occasion, the 27 member states and MEPs also accepted a European Commission proposal to postpone the entry into force of new rules intended to regulate so-called high-risk AI systems, those operating in sensitive areas such as security, health or fundamental rights.
Those rules had been due to take effect in August, but the Commission wanted to give businesses more time to adapt to this unprecedented framework.
The main change made to the Commission’s proposal by member states and MEPs was the adoption of fixed dates for the postponement, whereas Brussels had proposed a more flexible timetable.
Those dates are 2 December 2027 for “standalone” high-risk systems and 2 August 2028 for systems embedded in other programmes or products.
The agreement was reached as fears over the risks linked to artificial intelligence have been renewed in recent weeks in the EU by Mythos, the new model from the US start-up Anthropic, whose alleged capabilities in cybersecurity have fuelled concern.
Source: ANA-MPA.
