Evan HubingerAI alignment researcherPACESlow frontier capability development through verifiable compute allocation to safety research; this is not a blanket halt to AI training.“Here's a simple proposal for verifiable and enforceable slowdown that also differentially reallocates resources to alignment research”