Media Companies Seek New Rules to Protect Content from AI Training

Major media organizations are calling for new laws to protect their content from use by trainers of artificial intelligence (AI) tools.

主要媒体组织呼吁制定新法律，保护其内容不被人工智能 (AI) 工具培训师使用。

The request for new laws, or regulations, was contained in an open letter released earlier this month. The leaders of several large news and publishing organizations signed the document.

本月早些时候发布的一封公开信中包含了制定新法律或法规的要求。 多家大型新闻出版机构领导签署了该文件。

They include officials of the Associated Press (AP), Gannett and the News Media Alliance, which represents hundreds of media publishers. Representatives from Getty Images, the National Press Photographers Association and Agence France-Presse also signed the document.

他们包括美联社 (AP)、甘尼特和代表数百家媒体出版商的新闻媒体联盟的官员。 盖蒂图片社、国家新闻摄影师协会和法新社的代表也签署了该文件。

The organizations stated their support for the “responsible” development and deployment of AI systems. One AI tool, known as a “chatbot,” has demonstrated the ability to produce human-quality writing based on short, written commands. Such tools are also known as “generative AI” or “large language models.”

这些组织表示支持人工智能系统的“负责任”开发和部署。 一种被称为“聊天机器人”的人工智能工具已经证明能够根据简短的书面命令生成人类质量的写作。 此类工具也被称为“生成人工智能”或“大型语言模型”。

But the letter also expresses the need to develop regulations “to protect the content that powers” the increasing number of AI tools in development.

但这封信还表示，有必要制定法规，“保护驱动”越来越多正在开发的人工智能工具的内容。

Media companies are worried about AI developers using their published content without permission. The media content enjoys copyright protections and the companies want governments to enact rules to restrict unapproved use of their material.

媒体公司担心人工智能开发者未经许可使用他们发布的内容。 媒体内容享有版权保护，公司希望政府制定规则，限制未经批准使用其材料。

AI developers need huge amounts of data to train systems that aim to produce human-level results. Developers often collect this data from publicly available websites in a process called “scraping.”

人工智能开发人员需要大量数据来训练旨在产生人类水平结果的系统。 开发人员经常通过称为“抓取”的过程从公开网站收集这些数据。

The letter states that the process of scraping gives AI developers free use of their media content. They can then use this data to create language models that strengthen their AI tools and businesses.

信中指出，抓取过程使人工智能开发人员可以免费使用他们的媒体内容。 然后，他们可以使用这些数据来创建语言模型，以增强他们的人工智能工具和业务。

A few media companies have already signed licensing agreements with AI developers to give them permission to scrape their content. Last month, the Associated Press completed a deal with chatbot developer OpenAI to license the news agency’s large news story collection.

一些媒体公司已经与人工智能开发人员签署了许可协议，允许他们抓取其内容。 上个月，美联社与聊天机器人开发商 OpenAI 达成了一项协议，获得该通讯社大量新闻故事集的授权。

Other news organizations have taken steps designed to block developers from collecting their content. The New York Times, for example, recently changed its “Terms of Service” agreement to include new guidelines for AI.

其他新闻机构已采取措施阻止开发者收集其内容。 例如，《纽约时报》最近更改了其“服务条款”协议，以纳入新的人工智能指南。

The new policy requires AI developers to have written permission to use any of its content to train language models. The policy covers different forms of content, including written text, images and audio and video material.

新政策要求人工智能开发人员获得书面许可才能使用其任何内容来训练语言模型。 该政策涵盖不同形式的内容，包括书面文字、图像和音频和视频材料。

The open letter also describes another big problem with chatbots and other AI tools: the production of false or misleading information presented as truth. It calls on AI developers to build into their systems tools to prevent such falsehoods.

这封公开信还描述了聊天机器人和其他人工智能工具的另一个大问题：将虚假或误导性信息呈现为事实。 它呼吁人工智能开发人员在他们的系统中构建工具来防止此类谎言。

News and media content created by AI tools “can distort facts and leave the public with no basis to discern what is true and what is made up,” the letter states. It adds that many language models can also produce results that include long-standing biases against minority and underrepresented communities.

信中指出，人工智能工具创建的新闻和媒体内容“可能会歪曲事实，让公众没有基础辨别真假”。 它补充说，许多语言模型还可能产生包括长期存在的针对少数群体和代表性不足社区的偏见的结果。

Many news companies are currently experimenting with generative AI tools. They hope to learn how the tools can best serve the news production process. Several major news organizations announced last month they had teamed up with AI developers to find new tools to help reporters do their work.

许多新闻公司目前正在尝试生成人工智能工具。 他们希望了解这些工具如何最好地服务于新闻制作过程。 几家主要新闻机构上个月宣布，他们已与人工智能开发人员合作，寻找新工具来帮助记者完成工作。

The AP recently issued new guidelines for the use of AI tools across all departments at the news agency. The guidelines ban the use of such tools “to create publishable content and images for the news service,” the AP said in a story about its new rules.

美联社最近发布了该通讯社所有部门使用人工智能工具的新指南。 美联社在一篇有关其新规则的报道中表示，该指导方针禁止使用此类工具“为新闻服务创建可发布的内容和图像”。

However, the news agency said that it is also urging its employees to study the technology to learn how its use might improve their work within the rules.

不过，该通讯社表示，它还敦促员工研究该技术，以了解该技术的使用如何在规则范围内改善他们的工作。