OpenAI 为 DALL-E 3 图像生成器添加水印，助用户识别 AI 生成内容
人工智能（AI）图像生成器 DALL-E 3 已宣布将加入水印，以帮助用户识别使用 AI 生成的内容。
OpenAI 表示，该水印将出现在 ChatGPT 网站和 DALL-E 3 模型 API 生成的图像中。移动端用户将于 2 月 12 日起看到水印。
水印包含两个部分：不可见的元数据组件和可见的 CR 符号，后者位于每个图像的左上角。
OpenAI 此举旨在解决人们对 AI 生成内容的真实性和透明度的担忧。该水印将有助于防止未经授权使用 AI 生成的内容，并确保用户能够了解图像的来源。
水印的加入是 OpenAI 致力于负责任地开发和部署 AI 技术的最新举措。该公司此前已采取措施，防止 DALL-E 3 被用于创建虚假信息或有害内容。
DALL-E 3 是 OpenAI 开发的一种功能强大的图像生成器，它可以根据文本提示创建逼真的图像。该模型在 2023 年 1 月向公众发布，并迅速成为艺术家、设计师和内容创作者的热门工具。
然而，DALL-E 3 也引发了人们对 AI 生成内容的担忧。一些人担心 AI 生成的图像可能被用来传播虚假信息或冒充真实照片。
OpenAI 的水印旨在解决这些担忧，并确保用户能够识别和信任 AI 生成的内容。该水印将有助于促进透明度，并防止未经授权使用 AI 生成的内容。
英语如下：
**Headline:** AI-Generated Images Get Watermarks to Help Spot Authenticity
**Keywords:** AI watermark, DALL-E 3, authenticity detection
**Body:**
OpenAI is adding a watermark to its DALL-E3 image generator to help users identify AI-generated content.
The watermark will appear on images generated through the ChatGPT website and the DALL-E 3 model’s API, OpenAI said. Mobile users will start seeing the watermark on Feb. 12.
The watermark has two components: an invisiblemetadata component and a visible “DALL-E” logo in the bottom left corner of each image.
The move is part of OpenAI’s efforts to address concerns about the authenticity and transparency of AI-generated content. The watermark will help prevent unauthorized use of AI-generated content and ensure that users can understand the source of an image.
The watermark is the latest in a series of steps OpenAI has taken to responsibly develop and deploy its AI technology. The company has previously taken steps to prevent DALL-E 3 from being used to create misinformation or harmful content.
DALL-E 3 is a powerfulimage generator developed by OpenAI that can create realistic images from text prompts. The model was released to the public in January 2023 and has quickly become a popular tool for artists, designers, and content creators.
However, DALL-E 3 has also raised concerns about AI-generated content. Some worry that AI-generated images could be used to spread misinformation or impersonate real photographs.
OpenAI’s watermark is designed to address these concerns and ensure that users can identify and trust AI-generated content. The watermark will help promote transparency and prevent unauthorized use of AI-generated content.
【来源】https://www.ithome.com/0/749/371.htm
Views: 5