{
  "id": 862,
  "url": "https://arxiv.org/abs/2606.18424v1",
  "title": "A Variational Framework for LLM Generator-Regulator Games",
  "summary": "This paper develops a variational framework for regulated language generation. Starting from autoregressive token sampling, we derive the induced distribution over complete messages and relate it to an entropy-regularized Gibbs law. Regulation is modeled as an optimal discriminator whose convex-dual value is an f-divergence, and the generator-regulator interaction is formulated as a saddle-point problem. The framework applies to moderation, censorship, AI deception detection, compliance auditing",
  "authors": "Quanyan Zhu",
  "category": "research",
  "topics": "bias-fairness,regulation,transparency",
  "orgs": null,
  "regions": null,
  "published_at": "2026-06-16T19:19:22.000Z",
  "fetched_at": "2026-07-14T14:14:50.325Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/862",
  "original_url": "https://arxiv.org/abs/2606.18424v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}