Briefing Note for Policy Makers: 'Achieving the Three Non-Negotiables'
This briefing note summarises key policy recommendations from The Safe AI for Children Alliance's framework, Achieving the Three Non-Negotiables: A Policy Framework for Protecting Children from AI Harm.
Achieving the Three Non-Negotiables
Purpose: To propose a practical, multi-layer framework ensuring that AI systems cannot harm children through exploitation, manipulation, or self-harm promotion.
This briefing outlines a strategic policy framework that defines the essential protections and structural architecture required to realise them. It serves as a starting point for legislators and regulators to translate these principles into detailed technical and legal standards for enforcement.
The Three ‘Non-Negotiables’
AI must never:
Create fake sexualised images of children
Be designed to make children emotionally dependent
Encourage children to harm themselves
These are baseline protections – achievable and essential for child safeguarding in the digital age.
Proportionate to risk: The proposed measures are comprehensive because the risks to children are severe and novel. Just as aviation safety required building new regulatory institutions, AI safety for children requires a systematic approach.
The Proposed Framework
A four-layer policy model for practical implementation:
Foundation Model Standards: Independent pre-release safety testing by certified third parties for high- capability AI models; public reporting; ongoing audits.
Application-Level Rules: Bans on manipulative or exploitative design (especially those creating dependency); mandatory child-safe content filtering; transparency obligations.
Monitoring and Enforcement: Continuous oversight, mandatory harm reporting, rapid takedown mechanisms, strong whistleblower protections, and mandatory regulator-linked user reporting.
Age Assurance and Access Controls: Privacy-preserving methods to ensure age-appropriate access and parental oversight.
Immediate Policy Priorities
- Mandate independent testing of foundation models
- Require transparent public reporting of safety measures and incidents
- Introduce clear liability for companies whose AI systems harm children
- Enact strong whistleblower protections within AI law
- Mandate easily accessible, regulator-linked reporting mechanisms for all child-accessible AI platforms
- Begin international collaboration on enforcement and safety standards
Key Message: AI regulation that protects children is not a barrier to innovation – it is the foundation of responsible technological progress. The Three Non-Negotiables are minimum civilisational standards for the AI era.
End of Briefing Note
Download a copy of the briefing note here:
Download a copy of 'About The Safe AI for Children Alliance (SAIFCA)' here:
Download the full Policy Pack, with full policy details, here:
