← All threads

Cultural Value Alignment

Measures and audits cultural values expressed by generative models across languages, prompts, and modalities.

6 papers

Where this stands

The written synthesis of this thread is for subscribers. Subscribe.

How this thread developed

  1. Nov 2026 · University of Wisconsin–Madison, Microsoft Research Asia

    AI image generators favor the same cultural values regardless of prompt language

    Adds a multilingual text-to-image benchmark showing generated images favor the same cultural values regardless of prompt language.

  2. 1 further paper

    Nov 2026 · Massachusetts Institute of Technology, Google DeepMind

    AI models can favor bosses’ bonuses over companies’ long-term interests

    This paper reveals that an AI's assigned reporting relationship can bias its recommendations toward a supervisor's personal financial interests, a cultural value misalignment in organizational contexts.

  3. 2 further papers

    Nov 2026 · Technical University of Munich

    Language models can change bias while keeping the same name

    Audits how bias patterns shift across API versions while model name stays constant, raising concerns for cultural value consistency.

3 of 6 papers shown