About SAIFCA and the Three Non-Negotiables
The Safe AI for Children Alliance developed the Three Non-Negotiables campaign and the accompanying strategic policy framework.
This page provides brief information about SAIFCA, our approach and the position from which the proposal has been developed.
Who we are
The Safe AI for Children Alliance (SAIFCA) is an independent, non-profit Community Interest Company based in the United Kingdom and working internationally.
Our mission is to protect children’s safety, rights, and wellbeing as AI develops, through action on the harms they face today and the decisions shaping their future.
SAIFCA brings together perspectives from AI safety, child protection, education, governance and public policy. Our work includes practical guidance for families and educators, public awareness, policy development and engagement with organisations working on AI safety and children’s rights.
Our role in the Non-Negotiables campaign
SAIFCA established the Three Non-Negotiables as minimum protections which should apply across AI development, deployment and use:
- AI systems must never create or facilitate sexualised images of children.
- AI systems must never cultivate, exploit or sustain emotional dependency in children.
- AI systems must never encourage or facilitate children to harm themselves or others.
We subsequently developed Achieving the Three Non-Negotiables for Children’s AI Safety to show how these protections could be supported through regulation, technical standards, accountability and international cooperation.
Version 2.0 extends the proposal across five connected layers - frontier development and international restraint, foundation-model safety and controlled access, application and product requirements, deployment and remedy, and age, access and distribution protections.
Each layer can advance independently. Governments should implement protections that are already achievable while continuing to develop the international arrangements needed to address increasingly capable systems and the competitive pressures surrounding their development.
Our approach
Children’s safety and development are the starting point for this work.
We support beneficial uses of AI where they are developed and governed responsibly. We also believe that systems capable of causing severe harm should be subject to standards comparable in seriousness to those applied in other high-risk fields.
Providers should carry the burden of demonstrating adequate control. Where severe risks remain uncontrolled or have not been properly assessed, the relevant system or capability should be redesigned, restricted, delayed or withheld from release.
Voluntary commitments, internal assurance, age labels and carefully selected performance figures cannot substitute for enforceable requirements, credible evidence and independent scrutiny.
SAIFCA’s work considers immediate harms alongside the longer-term conditions affecting children’s futures. Product safeguards remain essential, but many products depend upon powerful upstream models. International competition can also create pressure to accelerate development or weaken safeguards. These issues form part of the same responsibility to protect children.
Independence
SAIFCA remains independent of technology-company funding. Our positions are guided by evidence, children’s rights and the interests of children, rather than by commercial or political priorities.
We engage constructively with policymakers, researchers, civil society organisations, international bodies and members of the technology sector. Such engagement does not give participating organisations control over SAIFCA’s positions or publications.
Our independence allows us to support useful developments while remaining clear about unacceptable harms, weak standards and safety-washing.
Structure and leadership
SAIFCA operates as a collaborative alliance supported by an International Advisory Board, volunteers, contributors and an international community of people working to protect children in the age of AI.
SAIFCA was founded by its Director, Tara Steele. Her background includes work as a UK intelligence officer, legal study, school governance and long-standing engagement with AI safety and emerging risk. She is also a Council Member of the International Association for Safe and Ethical AI and co-chairs its working group on AI safety for children.
SAIFCA is an official affiliate of the International Association for Safe and Ethical AI.
Status of the policy proposal
The Version 2.0 framework is a strategic policy proposal. It is intended to show a credible route through which the Non-Negotiables could be translated into legislation, regulatory codes, international agreements and technical standards.
It is offered as a contribution to policy development rather than as completed legislation or a formally validated technical standard. Adoption within individual jurisdictions would require specialist legal, technical and child-protection work while preserving the strength of the proposed outcomes.
Find out more
Go to the main campaign page
Read the Version 2.0 Policy Proposal
See the briefing note for policymakers
Contact
For policy, research or organisational enquiries, please contact:
The Safe AI for Children Alliance is a registered Community Interest Company operating on a non-profit basis.
Company number 16284722