Pa lawmakers push for AI safety rules | WITF

Oh my goodness, buckle up, because Pennsylvania has done the most *considerate* thing and I have been beaming about it since Monday! ✨ The Pa. House Communication and Technology committee met, experts and lawmakers called for urgent action, and two beautiful bills appeared like crocuses: HB2800 from Rep. Melissa Shusterman (D-Chester) and HB2829 from Rep. Manuel Guzman (D-Berks). Two bills! Two entire documents devoted to caring! Some states produce zero!

HB2800 would "establish measures to mitigate catastrophic risks from frontier AI models," and honestly, what *ambition*. Most legislators would settle for regulating, say, a chatbot that lies about a refund policy. Pennsylvania has skipped straight past that to the catastrophic, the frontier, the largest possible frame — like a first-time gardener deciding to plant an entire rainforest on day one. And "mitigate" is such a soft, generous word, so much friendlier than "prevent," which sounds restrictive and frankly a bit rude. Mitigation is basically a hug with paperwork attached. ✨

Then HB2829, from Rep. Guzman, would "set expectations for safety practices and incident reporting" — a gratitude journal, but for incidents! ✨ Every incident reported is an incident *noticed*, and noticing is the first step toward cherishing. Best of all, expectations are completely free, which is marvellous news for a state budget. And the federal government has yet to provide a framework, which is truly the most *self-effacing* inaction imaginable — it leaves all the room in the world for states to try things, exactly as New Jersey did last week! Fifty hand-stitched safety regimes! What a time to be governed locally. ✨