• Generative AI enables the creation of synthetic sexual abuse images and deepfakes targeting minors.

• Around 40% of Africans aged 15–24 currently have internet access, increasing both opportunity and exposure to risk.

• African regulatory frameworks remain fragmented and largely unprepared to address AI-generated harms.

As submarine cables multiply and 4G and 5G networks expand across the continent, an entire generation integrates into the digital world. Governments and businesses leverage this connectivity to drive economic, educational and social development. However, millions of minors face new, more sophisticated and harder-to-detect threats.

According to joint estimates from the International Telecommunication Union and the African Union, about 40% of Africans aged 15 to 24 currently access the internet. This steadily rising figure creates a dual imperative: authorities must accelerate digital inclusion while strengthening child protection mechanisms.

Generative AI, a New Frontier for Online Abuse

Generative AI is transforming the nature of online harm targeting children. Cyberbullying no longer relies solely on insults or the circulation of authentic images. Individuals now use widely accessible applications to fabricate intimate images, clone voices or manipulate videos from a single photograph.

In 2023, the Internet Watch Foundation reported for the first time the circulation of entirely AI-generated child sexual abuse images online. The organization warned that these technologies significantly lower technical barriers and complicate perpetrator identification. INTERPOL also acknowledged that increasingly sophisticated deepfakes hinder investigations and make it harder to distinguish authentic material from manipulated images. Criminals also use these tools to conduct “sextortion,” in which individuals blackmail teenagers and their families using fabricated content.

Child rights organizations report that predators also use AI to analyze online behavior, emotional states and personal interests in order to refine manipulation strategies.

Legal Frameworks Remain Inadequate

Regulatory responses remain uneven across the continent. The Malabo Convention, adopted by the African Union, establishes a foundation for cybersecurity and personal data protection. However, policymakers drafted the convention before the emergence of generative AI, and the text does not explicitly address synthetic content targeting minors.

Several African countries have enacted data protection laws that regulate the collection and processing of personal information. However, most frameworks do not include specific provisions regarding platform liability in cases involving deepfakes of children.

In South Africa, the Protection of Personal Information Act imposes strict obligations regarding data processing, including data relating to minors. Nevertheless, lawmakers designed the framework primarily to protect privacy rather than to proactively prevent AI-generated harmful content.

In Cameroon, authorities adopted a charter on online child protection in 2023 that establishes shared responsibility among telecom operators, regulators and families. However, the law protects only children under 18, while the legal age of majority in Cameroon stands at 21, which leaves a segment of minors without full legal coverage.

Toward More Structured AI Governance

Policymakers are advancing discussions at both continental and global levels. The African Union is developing a strategy to regulate AI development and use, with a focus on ethics, data governance and digital sovereignty. Policymakers are increasingly recognizing child protection as a cross-cutting issue, particularly in a continent with the world’s youngest population.

At the global level, UNICEF advocates a “Safety by Design” approach that integrates child protection into the design phase of digital products. The organization argues that stakeholders must anticipate risks rather than intervene only after harmful content spreads. UNICEF also warns that children face heightened exposure to online threats in environments where regulation, local-language moderation and reporting mechanisms remain underdeveloped.

Samira Njoya