Generative AI models are emerging as a versatile tool across diverse industries with applications in synthetic data generation computational art personalization of products and services and immersive entertainment Here we introduce a new privacy concern in the adoption and use of generative AI models that of coincidental generation Coincidental generation occurs when a models output inadvertently bears a likeness to a realworld entity Consider for example synthetic portrait generators which are today deployed in commercial applications such as virtual modeling agencies and synthetic stock photography We argue that the low intrinsic dimensionality of human face perception implies that every synthetically generated face will coincidentally resemble an actual person all but guaranteeing a privacy violation in the form of a misappropriation of likeness.

翻译：生成式人工智能模型正成为跨行业的通用工具，应用于合成数据生成、计算艺术、产品服务个性化以及沉浸式娱乐等领域。本文提出生成式AI模型在应用与使用中产生的新型隐私问题——"偶然生成"。偶然生成是指模型输出结果意外与真实世界实体产生相似性的现象。以当前商业应用中部署的虚拟模特经纪公司和合成图库为例，合成人像生成器生成的每张人脸都会因人类面部感知的低维特性而偶然与真实人物相似，这几乎必然导致肖像盗用形式的隐私侵犯。