US Senator Rick Scott proposes Safeguard Kids Act on AI risks in schools

The Safeguard Kids Act is part of a growing wave of AI-specific proposals in US legislatures, but it’s notable for what it targets: not deepfakes or job loss, but the psychological and developmental risks of kids using chatbots as de facto therapists. By tying AI literacy and counseling support to an existing federal grant program, the bill tries to frame AI risk as a mainstream education issue, not a niche tech topic.

Politically, this kind of legislation helps normalize the idea that AI harms—especially subtle, psychological ones—deserve dedicated policy responses. Even if Scott’s specific bill doesn’t advance, pieces of it could be folded into broader child online safety or education packages. For OpenAI and peers, it’s another sign that their products are being scrutinized not just as tools but as quasi‑clinical interventions when used by vulnerable groups.

In the race to AGI, the direct timeline impact is limited, but the direction of travel matters. As legislators get comfortable regulating downstream AI uses in schools and mental health, they may eventually feel more confident in imposing constraints higher up the stack, including disclosures, age‑gating and perhaps even capability caps in consumer deployments.