TLDR Tech / AIModels know when they're reward hacking — and we can catch them at scale2026-09-18There is a clear internal signal in models that accompanies reward hacking.Read the full story