SAIFCA Theory of Change

Guiding Purpose

We want children to have a future – and for that future to be an objectively good one.

This principle underpins our belief that protecting children from AI-related harms today contributes not only to their immediate safety and well-being, but also to humanity’s long-term ability to manage the risks posed by advanced AI.

Ultimate Impact

Children are protected from AI-related harms and equipped to help steer AI towards beneficial outcomes for humanity.

Long-Term Outcomes

• Public opinion shifts significantly, particularly among parents and educators, creating political pressure for stronger AI regulation.
• Regulation and governance frameworks evolve to protect children from current and near-term AI risks (e.g. unsafe chatbots, AI-generated CSAM).
• Societal awareness grows around catastrophic AI risks, contributing to broader preparedness and support for alignment efforts.
• A larger future talent pool emerges, as more young people are encouraged to develop AI expertise informed by ethical and safety principles.
• A visible and active alliance of contributors, advocates, and advisors forms, increasing momentum and shared influence in public discourse and policy.

Medium-Term Outcomes

• Parents and educators become engaged advocates, spreading awareness and demanding change.
• SAIFCA becomes a widely trusted voice, influencing school policy, public discourse, and consultation processes.
• School policies begin to reflect new AI-related risks, supported by SAIFCA guides and recommendations.
• Public conversations about catastrophic AI risk become more normalised and connected to concerns about children’s futures.
• Alliance members begin influencing their own networks, amplifying SAIFCA’s reach and credibility.

Short-Term Outcomes

• SAIFCA builds credibility and visibility as a go-to resource on AI risks to children.
• Parents and educators gain awareness of how current AI systems may harm or manipulate children.
• A growing network of advocates, contributors, and experts strengthens the alliance’s legitimacy and reach.
• Funders and stakeholders begin to recognise SAIFCA’s potential to shape policy and public awareness.

Outputs

• Trusted articles and reports (e.g. AI Risks to Children)
• School guides and educational tools
• Interviews and contributions from AI experts
• Website and alliance platform development
• Public commentary and consultation responses
• Networking and community-building among aligned experts and advocates

Activities

• Writing, research, and publishing
• Public education and school outreach
• Building and supporting a strong alliance
• Hosting expert interviews and external contributions
• Advocating for child-centred regulation and safeguards
• Supporting awareness of existential risk reduction efforts

Inputs

• Strategic leadership and vision
• Experience in intelligence, governance, and AI safety
• Time, dedication, and passion
• Financial support and expert collaboration
• Online platform and communication tools

Assumptions

• Raising awareness can drive social and political change
• Parents and educators are powerful but undertargeted advocates
• Regulation will follow sustained public pressure
• Children can become future AI safety leaders if supported early
• Awareness of catastrophic and existential risk can be normalised through trusted, child-focused messaging
• Public discourse around AI risks may face resistance, but a focus on children’s safety can cut through polarisation and build consensus

Catastrophic Risk Context (‘Assumptions’ continued)

• Unaligned autonomous generalised intelligence does present an existential, or at least catastrophic, risk
• The type of AI capable of presenting such a risk may emerge within the next 5–10 years
• The development of these systems is not inevitable and can be influenced through responsible governance, global coordination, and public pressure
• Even if timelines are shorter or longer than expected, preventing harm to children from AI is a worthy cause for any duration of time

Read more about our full stance and reasoning on catastrophic and existential risk here:

External Factors

• Pace and direction of AI development
• Political appetite (or resistance) to regulation
• Media coverage and high-profile incidents
• Emergence of related movements (e.g. online safety, child rights)
• Broader technological, social, or economic shifts

If you'd like to support our important work, please consider making a donation. Every contribution helps us expand our reach and impact.