Safetyism

In the AI Story

Safetyism operates through a characteristic three-move structure. First, identify a genuine risk — depth atrophy, fluent fabrication, the erosion of critical thinking. Second, implement a protection — prohibition, surveillance, AI detection software. Third, ignore or dismiss evidence that the protection produces harms exceeding the risk. The structure is robust because each individual step is defensible in isolation; the pathology only becomes visible when one examines the aggregate cost.

The framework's intellectual ancestry runs through Nassim Taleb's concept of antifragility — the property of systems that gain from disorder rather than merely surviving it. Children, Haidt and Lukianoff argued, are antifragile: they require challenge, friction, and even adversity to develop the capacities that make them robust adults. Safetyism treats them as fragile instead, and the treatment produces the fragility it was supposed to prevent. iGen's anxiety epidemic, documented by Jean Twenge, is the measurable outcome of a generation raised under safetyist assumptions.

The AI migration of the concept was explicit. The American Affairs Journal published "Beyond Safetyism: A Modest Proposal for Conservative AI Regulation" in August 2025, making the connection overt. Schools deploying AI detection software were executing safetyist logic — privileging the appearance of control over the reality of developmental benefit, optimizing for institutional liability rather than student growth. The scaffolded autonomy alternative required giving up the safety theater in exchange for the harder work of building genuine competence.

Safetyism is particularly seductive in AI policy because AI's failure modes are invisible. A child who falls from playground equipment feels the fall. A child who accepts a fluent fabrication from Claude feels nothing — the output is smooth by design. Institutions responding to this invisible risk reach naturally for the most visible intervention available: prohibition. The visibility serves an institutional function (demonstrating responsibility to parents and boards) that is independent of whether the prohibition produces developmental benefit.

Origin

The term was coined by Haidt and Lukianoff in a 2015 Atlantic essay and elaborated in The Coddling of the American Mind (2018), drawing on Skenazy's earlier documentation of overprotective parenting practices. The three authors' subsequent collaboration through Let Grow institutionalized the critique.

Key Ideas

Safety as sacred value. Safetyism treats safety not as one good among many to be balanced against competing goods, but as an absolute that overrides every other consideration.

Antifragility denied. The framework refuses to acknowledge that human beings require stressors to develop — a refusal that produces fragility at scale.

Institutional liability drives policy. Safetyist policies win not because they produce better outcomes but because they eliminate institutional risk, rewarding prohibition regardless of developmental cost.

Cognitive distortions embedded. Catastrophizing, emotional reasoning, and dichotomous thinking — the CBT targets of cognitive distortion — are institutionally encoded by safetyist policy.

Debates & Critiques

Defenders of stricter AI regulation argue that the safetyism critique conflates consumer technology regulation with educational philosophy, and that genuine product safety concerns (privacy, manipulation, developmental psychology) are not adequately addressed by "let the kids struggle." The Skenazy response is that the critique is not against regulation but against the substitution of prohibition for development — and that well-designed regulation preserves the child's opportunity to encounter the tool under supportive conditions.