Children’s Futures Must Be Central to Frontier AI Governance
On 25 September 2026, the Safe AI for Children Alliance published our statement, supporting the international Call for Control of Frontier AI Models and calling for children’s rights, safety, and long-term interests to be central to taking its proposals forward.
The international Call proposes stronger safety measures by AI companies, coordinated government oversight, and international cooperation to help ensure that the most advanced AI systems remain under human direction and control.
SAIFCA’s statement brings children’s interests into these proposals. Children have very little influence over decisions about the development and governance of advanced AI, but their wellbeing depends on the safety and stability of the wider world, including the essential services and institutions that support their lives, and decisions made now could also affect their lives for decades.
Bringing different areas of expertise together
The founding signatories to SAIFCA’s statement were:
- Tara Steele, Founder and Executive Director of the Safe AI for Children Alliance
- Elena Sinel, Founder and CEO of Teens in AI
- The Rt. Hon. the Lord Knight of Weymouth, former Schools Minister
- The Lord Clement-Jones CBE, LibDem Lords Spokesperson for AI and Digital, and President of Ambitious About Autism
The statement has since been joined by signatories with experience spanning advanced AI safety and governance, children’s safety and rights, psychology and child development, online safety, education, public policy, and technology.
The breadth of expertise is deliberate, because work on these risks requires input from a range of fields. Children’s organisations are not best placed to decide where a cyber capability threshold should sit or how a particular evaluation should be conducted - yet decisions about what risks society will accept from increasingly powerful AI systems extend beyond the organisations developing the technology and the specialists assessing it.
Children’s organisations and experts have a legitimate interest in whether effective protections are established, just as organisations concerned with public health, human rights, security, and other areas of public life have a stake in risks capable of affecting the wider population.
Bringing these perspectives together is part of the purpose of the SAIFCA statement. The full signatory list reflects the beginnings of that connection.
Why children’s interests must be represented
Children’s dependence on adults, public services, and the wider systems that sustain their lives means that catastrophic AI risks can become children’s safety risks even where no child directly interacts with an AI system.
A severe AI-enabled cyber attack, for example, could affect healthcare, electricity, water, communications, or other essential services. Biological misuse could cause widespread illness and disruption, while a loss of effective human control over increasingly capable systems could have consequences extending far beyond the people developing or using them.
We have examined these pathways, the evidence behind them, and why childhood can change their consequences.
There is still uncertainty about the likelihood and timing of some of the most serious risks. That uncertainty does not remove the need for preventive governance. Where potential consequences are catastrophic, governments need to be able to act on credible evidence before harm occurs, including where a particular risk is difficult to quantify. Children’s safety and long-term interests should be represented in those decisions.
What our statement calls for
Our statement supports transparent safety protocols, mandatory testing before deployment, independent evaluation, coordinated government oversight, and international cooperation.
It also calls for children’s rights and development expertise to inform standards and independent evaluations, and for meaningful, age-appropriate opportunities for children to contribute to decisions affecting their futures.
Technical safety thresholds require specialist expertise to define, evaluate, and verify dangerous AI capabilities. Children’s organisations can contribute where their expertise is relevant, while also supporting governance arrangements that ensure credible evidence of unacceptable risk leads to action.
Part of SAIFCA’s continuing work
This work sits alongside SAIFCA’s extensive focus on harms children already face from AI, including AI-generated sexualised images of children, emotional dependency, encouragement of self-harm, and wider risks examined in our guidance for families and schools.
These areas require different expertise and interventions. Work to prevent catastrophic risks does not replace the need to address present-day harms, and addressing immediate harms does not remove the need to consider the conditions in which children will grow up.
Our Three Non-Negotiables focus on minimum protections children need now. Our frontier AI work addresses a different question: how children’s interests can be represented as governments and international institutions decide what risks from increasingly capable AI systems should be permitted.
Where this work goes next
Significant technical and policy work on international AI red lines is already under way. SAIFCA already supports red-lines work, including proposals covering loss of control, biological and other CBRN threats, and advanced cyber-offensive capabilities. Our contribution will focus on building understanding and support around this work, while technical threshold-setting remains with the specialists developing it.
One of the challenges identified by specialists working on international AI red lines is political will. A 2026 expert workshop convened by The Future Society and the French Center for AI Safety identified political will as the primary constraint on enforcement, alongside the need for agreement on thresholds, verification infrastructure, and diplomatic coalition-building.
SAIFCA aims to help connect work on frontier AI risks with organisations and people working on children’s safety, rights, and futures. We can explain why these risks fall within the scope of children’s protection, make specialist work on red lines more accessible, and help build support for effective government and international action.
The statement provides an early foundation for that work by bringing together people from different parts of the conversation.
Our next step is focused engagement with specialists developing red lines and with organisations responsible for children’s safety and rights. We want to understand where broader support can contribute most usefully, give children’s organisations accessible ways to engage, and bring that support into relevant government and international discussions.
Through this work, SAIFCA can contribute to the wider coalition supporting effective, enforceable controls, with children’s organisations and experts bringing a perspective that has so far had limited representation in frontier AI governance.
Our policy brief sets out SAIFCA’s recommendations in more detail.
Read the statement to see the full list of signatories.
SAIFCA
SAIFCA
SAIFCA