近日，知名人工智能研究公司 OpenAI 宣布，其先进的图像生成器 DALL-E 3 将于近期开始为生成的图像添加水印。这一举措旨在帮助用户更加轻松地识别出由人工智能技术生成的内容，增强公众对于 AI 应用的认识和理解。
根据 OpenAI 的介绍，这些水印将基于来自内容来源和真实性联盟 (C2PA) 的标准，以保证其可靠性和有效性。从 2 月 12 日起，移动端用户在访问 ChatGPT 网站以及使用 DALL-E 3 模型 API 服务时，将能够看到这些新增的水印。
水印的设计由两个部分组成：一部分是不可见的元数据组件，它将嵌入到图像中，但不影响图像的显示效果；另一部分是可见的 CR 符号，这个符号将被放置在每个图像的左上角，以方便用户快速识别。
OpenAI 表示，加入水印是公司推动 AI 透明度和可追溯性的一项重要举措。当前，随着 AI 技术的快速发展，越来越多的图像、视频等内容是通过 AI 手段生成的。这使得公众、媒体以及内容创作者在判断信息来源时面临挑战。通过引入水印，用户可以更容易地区分 AI 生成的内容与传统创作的内容，从而有助于提高信息消费的质量和安全性。
此外，水印的引入也标志着 AI 技术在自我监管和规范方面迈出了新的一步。OpenAI 强调，这一举措将进一步促进 AI 技术的健康发展，并为构建可信赖的 AI 生态环境贡献力量。
未来，随着水印技术的不断改进和应用的深入，我们有理由相信，人工智能技术将在保障内容真实性、维护网络环境的健康发展等方面发挥更大的作用。
英语如下：
**News Title: OpenAI to Launch DALL-E 3 Image Watermarking Feature**
Keywords: AI Image Watermarking, DALL-E 3, Content Authenticity Protocol Alliance (C2PA)
### OpenAI Adds Watermarks to DALL-E 3 Image Generator to Help Users Identify AI-Generated Content
Recently, OpenAI, a renowned artificial intelligence research company, announced that its advanced image generator, DALL-E 3, will begin to add watermarks to the generated images in the near future. This move aims to help users more easily identify content generated by artificial intelligence technology, enhancing public awareness and understanding of AI applications.
According to OpenAI, these watermarks will be based on the standards from the Content Authenticity Protocol Alliance (C2PA) to ensure their reliability and effectiveness. Starting February 12, mobile users visiting the ChatGPT website and users of the DALL-E 3 model API service will be able to see these new watermarks.
The design of the watermark consists of two parts: one is an invisible metadata component that will be embedded in the image without affecting the display of the image; the other is a visible “CR” symbol that will be placed in the upper left corner of each image to facilitate quick identification by users.
OpenAI stated that introducing watermarks is an important step in the company’s efforts to promote AI transparency and traceability. Currently, as AI technology rapidly advances, an increasing number of images, videos, and other content are generated through AI means. This challenges the public, media, and content creators in distinguishing information sources. By incorporating watermarks, users can more easily differentiate AI-generated content from traditionally created content, thereby helping to improve the quality and security of information consumption.
Furthermore, the introduction of watermarks signifies a new step in the self-regulation and standardization of AI technology. OpenAI emphasized that this initiative will further promote the healthy development of AI technology and contribute to the construction of a trustworthy AI ecosystem.
In the future, as watermarking technology continues to improve and its applications deepen, there is reason to believe that artificial intelligence technology will play an even greater role in ensuring content authenticity and maintaining a healthy online environment.
【来源】https://www.ithome.com/0/749/371.htm
Views: 9