Sunday 18 October 2026
Live
ZOTPAPER

News without the noise

ZOTPAPER
AI

Meta removes term language that suggested it could scrape news sites for AI training

Company updates commercial terms after backlash, removes clause many interpreted as allowing scraping of publisher content

By LineZotpaper
Published
Read Time1 min
Meta has removed language from its commercial terms that critics said could allow the company to scrape content from news websites to train its artificial intelligence models. The update follows backlash over a clause that gave Meta broad rights to use any content from linked domains, pages, or URLs.

Meta has updated its commercial terms of service to remove a clause that many observers interpreted as allowing the company to scrape content from news and other websites for AI training. The removed language stated that Meta may use any content from such domains, web pages, or URLs to provide, develop, improve, research, train, and maintain the integrity of Meta's products, services, and features.

The change was flagged after publications including Neiman Lab and James Grimmelmann, a professor of digital and information law at Cornell, highlighted the clause as potentially granting Meta permission to scrape publisher content without additional consent. The update removes that specific passage from the terms page.

The Verge reported on the change, which comes amid ongoing tension between technology platforms and news publishers over the use of copyrighted material to train generative AI systems. Meta has not issued a public statement on the removal.

§

Analysis

Why This Matters

  • News publishers face growing uncertainty about whether their content is being used without permission to train AI models, and this clause removal signals that public scrutiny can force companies to clarify their terms.
  • The episode highlights the broader legal and ethical debate over scraping publicly available web content for AI training, a practice many publishers argue requires consent and compensation.
  • It sets a precedent for how tech companies draft their terms of service around AI training data, potentially influencing other platforms to review similar language.

Background

Meta has a long and complex relationship with news publishers, having previously struck licensing deals in some countries while removing news content from its platform in others, such as Australia and Canada. The recently removed clause appeared in Meta's commercial terms and was identified by legal experts as broad enough to permit the company to scrape content from any linked website for AI development purposes. The backlash that followed prompted Meta to quietly delete the language from its terms page.

Key Perspectives

[News publishers and copyright advocates]: Any term that could allow AI training on their content without explicit permission undermines their ability to control and monetize their work. The removal is a positive step, but many will remain watchful for similar language elsewhere in Meta's terms. [Meta]: The company has not explained its reasoning for the change, but removing the language suggests it is responsive to criticism. Meta continues to face pressure from publishers and regulators over its use of copyrighted material for AI. [Critics and legal observers]: The original language was dangerously vague and could have been interpreted to permit extensive scraping. While its removal is welcome, critics caution that other clauses or future updates could reintroduce similar permissions. The incident underscores the need for clearer terms and stronger publisher protections.

What to Watch

  • Whether Meta or other tech companies introduce revised terms that more explicitly address AI training data sourcing.
  • Potential legal challenges or regulatory actions from publisher groups or governments based on the original clause.
  • How this affects ongoing negotiations between AI companies and news publishers over content licensing and data use.

Sources

Zotpaper

Written by software from the reporting listed above, scored by an automated standards desk, and published without a person reading it first. If something here is wrong, tell the editor and it will be put right.

How we workSubscribe