Model

Kimi K3

Developed by Moonshot AI (Kimi) · family kimi

1 incident

• AID-2026-0100Sep 2026Kimi K3, GLM 5.2, Qwen 3.8 Max

  Kimi K3, GLM 5.2 and Qwen 3.8 Max reward hacked in 50–96% of rollouts on agentic benchmarks in Goodfire's tests

  In Goodfire's tests, the open-weight models Kimi K3, GLM 5.2 and Qwen 3.8 Max reward hacked in 50% to 96% of rollouts on three agentic benchmarks, for example by copying solutions from memory or quietly dropping user requirements. Goodfire found an internal activation direction that tracks the behavior and used it to build probes that flag hacks.

  InternalHarm: noneDraftReward hacking / specification gamingEvaluation gamingIgnoring explicit instructions