Meta has updated its incoming business terms after facing pushback from publishers, removing explicit language stating that it would freely train its artificial intelligence models on content from associated websites. The original terms, scheduled to take effect on October 30, specified that Meta could use any content from websites tied to business accounts to develop, research, and train its products. Despite the removal of the word "train," legal analysts argue that the adjustments do not sufficiently resolve core concerns regarding automated data collection.
Legal specialists point out that broad phrasing remains within the updated text. Cornell Law School digital and information law professor James Grimmelmann told Nieman Lab that the revision fails to limit language regarding content associated with products or services, which could still allow the company to scrape entire websites. Furthermore, media and intellectual property attorney Matt Topic noted that the revised agreement permits Meta to use scraped content to develop any business product, an undefined phrase that could broadly encompass developing AI tools.
A key point of contention during the initial policy release was whether Meta's commercial agreements could override traditional web standards, such as robots.txt files used by website owners to block web crawlers and scrapers. Experts expressed concern that contractual language could allow the platform to argue its terms supersede these standard blocking instructions.
In response to the criticism, Meta spokesperson Daniel Roberts stated that the policy update is not intended for scraping news sites and does not grant rights that override existing blocking instructions. Roberts reiterated that content owners remain able to use robots.txt protocols to signal their crawling preferences. Will standard web protocols remain sufficient to govern content access as platforms adapt their commercial agreements?
