UNICEF Guidance: Ensuring Children's Rights in the Age of AI

The foundation for all policies, laws, and decisions affecting children, including those related to AI, rests upon the UN Convention on the Rights of the Child (CRC). This foundational document, ratified by nearly every country, asserts that children's rights apply equally in the digital environment. Child-centred AI is defined as AI that ensures the fulfillment and protection of all children's rights, encompassing systems that may indirectly impact them. States have a duty to protect children's rights in the context of AI, while businesses bear the responsibility to respect these rights both offline and online, guided by principles like the UN Guiding Principles on Business and Human Rights and UNICEF's Child Rights and Business Principles. Child-centred AI is not an option but an obligation to future generations. The CRC's principles of Protection (do no harm), Provision (do good), and Participation (include all children) are paramount. Protection ensures children are shielded from harmful AI impacts and can interact safely. Provision means leveraging AI opportunities for education, health, and play when safe and beneficial. Participation empowers children to shape AI systems and make informed decisions about their use and impact. Crucially, AI policies and systems must be age-appropriate, recognizing the evolving capacities of children and adapting to developmental stages, ensuring that all children under 18 benefit from CRC rights, regardless of digital consent age.

Ensuring the safety of children in the context of AI is a paramount requirement. This involves proactively identifying and mitigating potential harms that AI systems can pose to children, both directly and indirectly. Key concerns include AI-generated child sexual abuse material (CSAM) and non-consensual intimate images (NCII), which require stringent measures for prevention, detection, and removal. Furthermore, AI systems must be designed to prevent exposure to harmful content, such as hate speech, disinformation, and content promoting self-harm or dangerous activities. The guidance emphasizes the need for AI systems to be safe for children's interaction, avoiding addictive design patterns, emotional manipulation, and the creation of unhealthy dependencies, particularly with AI companions or chatbots. This also extends to ensuring the safety of children involved in the AI supply chain, addressing issues like child labor and the use of contaminated datasets. Robust safety protocols, risk assessments, and continuous monitoring are essential to protect children from the myriad risks associated with AI.

AI systems must be designed and deployed to ensure non-discrimination and fairness for all children, regardless of their background, identity, or circumstances. Algorithmic bias, often stemming from biased training data or flawed design, can perpetuate and amplify existing societal inequalities, leading to discriminatory outcomes in areas such as education, healthcare, and justice. This requirement calls for rigorous testing and auditing of AI systems to identify and mitigate biases that could disadvantage children based on race, gender, disability, socioeconomic status, or other protected characteristics. Efforts should be made to ensure AI systems are inclusive and equitable, providing equal opportunities and access to benefits for all children. This includes addressing disparities in AI access and usage, particularly for children from marginalized communities or the Global South, ensuring that AI development does not exacerbate existing digital divides.

Responsible AI practice necessitates a deep commitment to respecting human rights and, specifically, children's rights throughout the entire AI lifecycle. This means embedding human rights principles into the design, development, deployment, and ongoing use of AI systems. For businesses, this aligns with the UN Guiding Principles on Business and Human Rights and UNICEF's Child Rights and Business Principles, requiring due diligence to identify, prevent, mitigate, and account for how AI impacts human rights. For governments, it means ensuring that AI policies and regulations are consistent with their international human rights obligations, including the CRC. This requirement also extends to addressing the broader societal impacts of AI, such as its role in armed conflict and cyber operations, ensuring that AI is used in ways that uphold peace, security, and human dignity, and does not undermine fundamental rights.

Ensuring the inclusion of and for children in the AI ecosystem is fundamental to child-centred AI. This means actively involving children in the design, development, testing, and governance of AI systems that affect them. Their perspectives, experiences, and needs must be central to these processes. This is particularly important for children from marginalized or vulnerable communities, including those in the Global South, rural areas, and children with disabilities, who are often excluded from policy forums and AI design processes. Participatory research methods and child-led consultations are essential to gather diverse insights and ensure that AI solutions are relevant and beneficial to all children. Inclusion also means ensuring equitable access to AI technologies and the opportunities they provide, bridging digital divides and empowering all children to benefit from AI advancements.

Creating an enabling environment for child-centred AI requires a concerted effort from all stakeholders to foster collaboration, innovation, and responsible governance. This involves promoting interdisciplinary research to better understand the complex interactions between AI and children, and sharing emergent evidence widely. Governments and businesses need to work together to develop interoperable governance frameworks that can adapt to the rapid pace of technological change. International cooperation is essential to address the global nature of AI and ensure consistent protection for children across borders. This also includes investing in capacity building for policymakers, developers, and educators, and supporting civil society organizations in their role of monitoring and advocating for child-centred AI. Ultimately, an enabling environment is one where children's rights are consistently prioritized, and AI is harnessed as a force for good in their lives.