Evil individuals creating sexual deepfakes of children, even infants younger than two years old, will be targeted by a new law enforcement initiative. The legislation allows AI developers and child protection groups to test artificial intelligence models to prevent the production of inappropriate images.
Presently, UK law prohibits the possession and distribution of child sexual abuse material but restricts developers from conducting safety evaluations on AI models. Consequently, harmful images can only be addressed after they have been circulated online.
According to the Internet Watch Foundation (IWF), cases of AI-generated child sexual abuse material have surged in the past year, from 199 in 2024 to 426 in 2025. Notably, there has been a disturbing increase in depictions involving infants, with images of 0–2-year-olds rising from five to 92 in the same period.
The severity of the content has also escalated, with Category A material, encompassing explicit sexual activities, rising from 2,621 to 3,086 items, predominantly targeting girls. In a groundbreaking move, a modification to the law will enable rigorous testing of AI systems’ safeguards from the outset, ensuring protection against extreme content and non-consensual imagery.
This amendment to the Crime and Policing Bill will convene a panel of AI and child safety experts to oversee the secure implementation of these changes. The NSPCC stresses that mandatory testing of AI models is imperative to combat child sexual abuse effectively.
Rani Govender, policy manager at the charity, emphasized the necessity of mandatory provisions for AI developers to prioritize child safety in product design. Kerry Smith, CEO of the IWF, highlighted the potential of AI tools to perpetuate victimization and the importance of integrating safety measures into new technologies.
Technology Secretary Liz Kendall affirmed the government’s commitment to prioritizing child safety in the face of technological advancements. The new laws aim to embed safety features into AI systems to proactively mitigate risks to children.
By empowering trusted organizations to scrutinize AI models, the legislation ensures that child safety is a fundamental consideration in AI development. This proactive approach seeks to prevent vulnerabilities and protect children from potential harm posed by emerging technologies.