Abstract:
On October 17, local time, Meta revised its commercial service terms that were about to take effect, deleting a statement that had previously caused strong concern among media publishers and legal experts. The original terms clearly mentioned that Meta can use content obtained from relevant websites, web pages or URLs to develop, improve, research and train its products and services. Since this wording may involve the content of third-party websites such as news websites, the outside world is worried that Meta is trying to use commercial service agreements to provide a basis for large-scale crawling of Internet content and use it for artificial intelligence training.
The changes come after public criticism of the terms in question. Previously, when Meta updated its commercial terms of service, it linked access to websites and content through automated programs to the development and training of company products. Since a large number of news media, publishing organizations and enterprises have commercial accounts on Facebook, Instagram and other platforms, these organizations are worried that as long as they continue to use Meta's commercial services, they may be regarded as accepting new terms, thereby giving Meta broad authorization to use the content of its associated websites.
The core of the dispute is that the relevant content is not limited to posts actively posted by users on Facebook or Instagram, but may extend to external websites associated with commercial accounts. For news outlets, this means articles, images and other content from their official websites may be included in the discussion. Critics worry that Meta may use this to bypass traditional content authorization and licensing negotiations and use web scraping technology to obtain large amounts of content without reaching individual agreements with publishers.
Nieman Journalism Lab has previously reported on this issue and interviewed relevant legal experts. Cornell University law professor James Grimmelman and others believe that the terms originally announced by Meta were quite broad in wording and may have been understood to allow companies to crawl website content related to commercial accounts and use them for artificial intelligence training. This has raised concerns in the media industry about copyright protection, content licensing and Internet regulations.
Representatives from digital media industry groups also expressed dissatisfaction. They pointed out that the establishment of commercial partnerships between news organizations and social platforms does not mean that the platforms should automatically obtain the right to use content on their official websites. If the relevant terms are interpreted broadly, Meta may be able to obtain commercially valuable news and published content without paying separate content licensing fees. This may not only affect the existing content licensing business of media organizations, but may also change the profit distribution relationship between artificial intelligence companies and content producers.
In the face of external doubts, Meta has previously denied that its web crawling behavior has changed. Company spokesman Andy Stone has said that claims that Meta will change the way web crawlers are used are untrue. Meta emphasized that the relevant terms mainly involve advertising and commercial services, and are not intended to capture the content of news websites or expand the scope of access to artificial intelligence training data.
In the latest revision, Meta has deleted the clear statement in the original terms that allows the use of relevant website content for the development, improvement, research and training of products, and further clarified that the relevant web scraping terms only apply to advertising and commercial products.
A Meta spokesperson said that the company has listened to outside opinions and hopes to eliminate misunderstandings by amending the terms. This adjustment only further clarifies the scope of application of the original terms and does not mean that Meta has changed the way web crawlers are used. The company also emphasized that website owners can still express their wishes for automated access through the robots.txt file.
Robots.txt is a widely used website control mechanism. This file allows website operators to declare to search engines and other automated access programs which pages are allowed to be crawled and which pages they do not want to be accessed. While robots.txt is not equivalent to a copyright license and does not necessarily have the same legal effect in all legal contexts, it has long been an important technical tool for website administrators to express crawling rules.
Therefore, Meta’s revision of the terms this time does not just delete a word involving artificial intelligence. It also touches on issues such as platform commercial agreements, the control that website owners have, and the rules that automated programs should follow when accessing publicly available Internet content.
However, the revised terms have not completely eliminated industry concerns. Some media industry representatives believe that although Meta has deleted text that explicitly mentions artificial intelligence training, this does not necessarily mean that the company has completely ruled out the possibility of using scraped content for artificial intelligence-related purposes. The new formulation still relates to advertising and commercial products, which themselves may also use artificial intelligence technology.
In addition, there is room for further discussion on how to define the boundaries between advertising and commercial products and artificial intelligence training. If the information obtained from the website is used for product recommendation, advertising content generation or other commercial functions, then whether these uses may indirectly involve the development and improvement of artificial intelligence models still needs to be judged based on specific terms and actual operations.
The controversy is also related to the increasingly fierce data competition in the artificial intelligence industry. Big tech companies are investing heavily in developing generative AI products, and high-quality data is critical for training, optimizing, and evaluating models. News websites, professional publications and corporate official websites have accumulated a large amount of edited, organized or professionally created content, which has significant information value. How to obtain this content, and whether licensing and remuneration are required, have become important legal and business issues facing the artificial intelligence industry.
Meta has continued to expand its artificial intelligence business in recent years, developing consumer-oriented AI products and integrating artificial intelligence technology into advertising, content recommendations and business services. For companies, accurately obtaining product information on third-party websites does help improve advertising and business recommendations; but for website operators, allowing automated programs to obtain product information is not the same thing as allowing them to crawl news articles, copy original content, and use it for artificial intelligence training.
This also explains why the initial wording of the clause triggered such a strong reaction. Even if Meta claims that its purpose is only to support advertising and commercial services, if the contract text uses expressions covering a wide range of purposes such as development, research, and training, outsiders may still worry that these permissions exceed the reasonable expectations of commercial partners.
This modification occurs as Meta’s new version of the business terms is about to take effect. According to the previously announced arrangements, the relevant provisions are expected to take effect on October 30. Meta has tried to reduce the concerns of publishers and corporate customers by deleting controversial text and clarifying the scope of application, but it remains to be seen whether the revised terms can fully respond to outside concerns about content licensing and artificial intelligence training.
From a broader industry perspective, this controversy reflects that Internet content usage rules are undergoing new adjustments. In the past, websites mainly faced search engines, advertising systems, and traditional web crawlers; now, artificial intelligence training, generative search, and AI agents that can perform tasks on behalf of users have brought new uses for automated access. Website owners not only need to consider whether their content can be indexed by search engines, but also whether different types of programs should be allowed to access, copy and use their content.
For news publishers and other content producers, the real question is not just whether a company removed a word from a contract, but how to ensure that website content remains subject to clear, enforceable usage rules after it is captured by automated systems. For Meta, this revision shows that in the context of the growing demand for artificial intelligence data, technical wording in enterprise terms of service may also trigger widespread copyright and commercial disputes.
For now, Meta insists that the change does not change how its web crawler is actually used, nor does it give the company new rights to override website owners' existing crawling preferences. But publishers and legal experts are still concerned about how much protection the revised terms will provide. As artificial intelligence companies increasingly rely on online content, such disputes surrounding data access, copyright licensing and platform contracts are expected to continue to become important issues in the technology industry.
Comments