The European Union has reached a provisional agreement to update its criminal law rules on child sexual abuse, including offences involving artificial intelligence, livestreaming and sexual extortion.
The agreement, reached by the Council presidency and European Parliament negotiators on 22 June, expands the EU’s legal framework against child sexual abuse and child sexual exploitation. The provisional deal will require formal approval by both institutions before becoming law, but it marks a further step in the EU’s attempt to adapt criminal law to offences increasingly enabled by digital tools.
The revised rules would cover a broader range of offences, increase penalties and extend limitation periods, giving prosecutors more time to bring cases. The text also strengthens provisions on victim support and sets clearer rules on consent.
One of the main changes concerns AI-generated child sexual abuse material. The rapid development of generative AI has created new problems for investigators, prosecutors and online safety bodies. Synthetic material can be created without an original abuse image, altered from existing photographs, or used to simulate abuse involving identifiable children. This complicates detection, evidence-gathering and prosecution.
EU policymakers have been under pressure to close gaps in the existing framework. Current rules were designed before the spread of widely accessible generative AI tools capable of producing realistic images and videos. A recent European Parliament research note on AI image generation and online child sexual abuse said that reports of AI-generated child sexual abuse material had risen sharply, raising questions over how law enforcement and regulators should respond.
The new agreement also addresses livestreamed abuse and sexual extortion. These offences can cross jurisdictions, involve encrypted communication and leave limited physical evidence. Victims may also face threats that images or videos will be distributed unless they comply with further demands.
The legal update is separate from the EU’s wider and more controversial debate on combating child sexual abuse online. That debate has divided governments, technology companies, privacy groups and child-protection advocates over the extent to which online services should be required or allowed to detect and report suspected material. The current agreement focuses on criminal law definitions, penalties and victim rights rather than on platform-monitoring obligations.
The distinction matters. Criminalising conduct is different from deciding how private companies, messaging services and hosting providers should detect and report suspected material. The EU is therefore moving on two tracks: updating the criminal-law framework while still struggling to settle the rules for online detection.
For member states, the practical question will be implementation. Criminal law remains primarily enforced at national level. Once the directive is formally adopted, governments will need to ensure that police, prosecutors and courts have the legal tools and technical capacity to investigate offences involving AI-generated material, livestreaming and sexual extortion.
That may require more than new statutory wording. Investigators will need forensic tools capable of assessing manipulated or synthetic material. Prosecutors will need to establish intent, possession, distribution and production in cases where the content may not originate from a conventional camera image. Courts may also have to deal with questions over identification, authenticity and evidential standards.
The agreement also reflects a wider change in EU digital policy. Brussels has increasingly moved from treating online harms only as a platform-governance issue to treating them also as a criminal-law and security issue. AI-generated abuse material sits at the intersection of both. It involves the misuse of technology, but also creates direct legal consequences for victims, offenders, platforms and developers.
The EU has already moved to restrict certain harmful AI practices through changes to its artificial intelligence rules. In June, the European Parliament approved measures under the AI Act banning AI systems that generate child sexual abuse material or create non-consensual intimate images, video or audio involving identifiable people. The criminal-law agreement adds another layer by focusing on offences, penalties and prosecution.
The measure is likely to be broadly supported in principle, but its effectiveness will depend on enforcement. Synthetic material can be generated quickly, shared across borders and moved between platforms. It may also be hosted outside the EU, requiring cooperation with third countries and technology providers.
For Brussels, the agreement allows the bloc to present itself as updating child-protection law for the AI era. For national authorities, the harder test will be whether the new rules lead to more effective investigations and prosecutions rather than only broader legal definitions.
The provisional deal is therefore important not because it completes the EU’s response to online child abuse, but because it confirms that AI-generated material is now being treated as part of the criminal-law problem. The next stage will show whether member states can turn that legal recognition into practical enforcement.
