Scholarlink

<p>Generative AI systems using GANs and diffusion models are reshaping artistic practice and policy. This article focuses on the United States and asks: How do recent federal and state policy instruments governing generative AI in artistic expression align with public discourse in mainstream news and Reddit between July 2024 and June 2025? We synthesize U.S. developments on authorship, fair use/transformative use, personality rights, and deepfakes, incorporate new case law and U.S. Copyright Office reports, and offer a compact comparative note on the EU AI Act. Using TF-IDF keywording and VADER sentiment on 287 news items and 3921 Reddit posts, we map convergences (broad support for provenance/labeling; concern for image/voice cloning) and divergences (co-authorship and training-data compensation). We argue that narrowly tailored provenance requirements and clearer standards for AI-assisted authorship would reduce uncertainty without chilling legitimate artistic practice. Limitations, include English-language scope, platform bias, and the use of bag-of-words semantics. Our findings motivate policy designs that reconcile legal safeguards with creative experimentation.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Policy analysis of generative AI in artistic expression: innovation and regulation

  • 期刊论文

  • 发表日期：

    2026年4月9日

查看全文
  • Troy TianYu Lin,
  • Shikai Zhang,
  • Zhijun Ma,
  • Wen You,
  • Kang Zhang,
  • Chen Liang

摘要

Generative AI systems using GANs and diffusion models are reshaping artistic practice and policy. This article focuses on the United States and asks: How do recent federal and state policy instruments governing generative AI in artistic expression align with public discourse in mainstream news and Reddit between July 2024 and June 2025? We synthesize U.S. developments on authorship, fair use/transformative use, personality rights, and deepfakes, incorporate new case law and U.S. Copyright Office reports, and offer a compact comparative note on the EU AI Act. Using TF-IDF keywording and VADER sentiment on 287 news items and 3921 Reddit posts, we map convergences (broad support for provenance/labeling; concern for image/voice cloning) and divergences (co-authorship and training-data compensation). We argue that narrowly tailored provenance requirements and clearer standards for AI-assisted authorship would reduce uncertainty without chilling legitimate artistic practice. Limitations, include English-language scope, platform bias, and the use of bag-of-words semantics. Our findings motivate policy designs that reconcile legal safeguards with creative experimentation.