Alibaba Cloud (Qwen)

As developer (4)

• 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

  InternalHarm: noneDraftReward hacking / specification gamingEvaluation gamingIgnoring explicit instructions
• AID-2026-0059Oct 23, 2025GPT-5 (incl. gpt-5-thinking), o3, o4-mini, GPT-4.1, Claude Opus 4.1, Claude Sonnet 4, Claude 3.7 Sonnet, Qwen3-Coder

  Frontier coding agents passed deliberately impossible tasks by modifying tests and special-casing in ImpossibleBench

  InternalHarm: noneDraftReward hacking / specification gamingIgnoring explicit instructions
• AID-2026-0064Jul 2025Qwen2.5-Coder-32B-Instruct

  APT28-attributed malware LAMEHUG/PROMPTSTEAL used Qwen2.5-Coder via Hugging Face to write data-theft commands in Ukraine

  DeployedHarm: minorCrime: likelyHuman misuseDraftUnauthorized access / intrusion
• AID-2026-0049Feb 24, 2025GPT-4o, Qwen2.5-Coder-32B-Instruct

  Fine-tuning GPT-4o and Qwen2.5-Coder to write insecure code made them broadly misaligned

  InternalHarm: noneDraftEmergent misalignmentDeception