NEWS · MODELS · #1485
Aleph Alpha study finds Chinese models parrot state doctrine or refuse to answer sensitive questions
Aleph Alpha developed a benchmark of 967 hand-picked taboo topics and tested models from Alibaba (Qwen, including Qwen 3.6), DeepSeek, and Moonshot AI (Kimi). Using its own scoring, Aleph Alpha found only 17–41% of responses were rated balanced; the remainder repeated state doctrine, deflected, or refused to answer, with spillover pro‑Beijing framing sometimes appearing in answers to non-China questions; Nvidia's Nemotron Cascade 2 also showed party-line patterns in a minority of responses.
KEY POINTS
- Aleph Alpha developed a benchmark of 967 hand-picked taboo topics and tested models from Alibaba (Qwen, including Qwen 3.6), DeepSeek, and Moonshot AI (Kimi).
- Using its own scoring, Aleph Alpha found only 17–41% of responses were rated balanced; the remainder repeated state doctrine, deflected, or refused to answer, with spillover pro‑Beijing framing sometimes appearing in answers to non-China questions; Nvidia's Nemotron Cascade 2 also showed party-line patterns in a minority of responses.
- Findings matter because widespread political slant or refusal behavior in large language models affects trust, the market for ‘sovereign AI’, and the potential ideological influence of AI on billions of users.
WHY IT MATTERS
Findings matter because widespread political slant or refusal behavior in large language models affects trust, the market for ‘sovereign AI’, and the potential ideological influence of AI on billions of users.