Meta's Revised Business Terms Still Leave Publishers Exposed to AI Use
Meta dropped the word 'train' from its new business terms, but lawyers say AI use of scraped news remains possible. Here is what it means for Indian publishers.
The News
Meta has revised the business terms it rolls out on October 30th, deleting wording that appeared to let it train its models on content pulled from websites linked to business accounts. Legal specialists say the edit falls short, because the remaining text still lets Meta use that material to build products.
The original draft said Meta could use content from a business account's associated websites to research, develop and train its products. After strong objections from publishers, the word "train" was dropped. Matt Topic, a media and intellectual property attorney at Loevy & Loevy, told Nieman Lab that the surviving permission to develop an undefined "business product" is wide enough to cover training anyway.
James Grimmelmann, professor of digital and information law at Cornell Law School, told the same outlet that the clause covering sites "associated with your products or services" still leaves the door open to scraping whole websites. Meta disagrees. Spokesperson Daniel Roberts told The Verge that "content owners can use robots.txt to signal their preferences", and said the update grants Meta no new rights that override them.
Why It Matters
The dispute is less about one clause and more about which rulebook governs AI crawling: a voluntary file such as robots.txt, or a contract a publisher accepts simply by running a business page. Lawyers quoted in the original reporting warned that commercial terms could be argued to sit above crawler-blocking instructions. If that reasoning held, every publisher with a Facebook or Instagram business presence could find its opt-out weakened by an agreement it signed for advertising, not licensing.
The pattern is familiar. When The New York Times sued OpenAI and Microsoft in December 2023, the core question was whether scraped journalism could be used for training without payment. Since then, publishers have split between licensing deals and litigation, with terms-of-service drafting a quieter third front. Meta's retreat on one word shows that publisher pressure works, but also that broad catch-all language can achieve much the same result.
Indian Angle
India already has a live test of these questions. News agency ANI sued OpenAI in the Delhi High Court in November 2024, alleging its content was used for training without permission, and the Digital News Publishers Association sought to join the case. A ruling there will shape how Indian courts read the Copyright Act, 1957 against AI training, and whether a robots.txt opt-out is treated as a meaningful signal of consent.
Meta's reach in India makes the business-terms question practical rather than theoretical. Most Indian newsrooms, from national dailies to regional-language digital outlets, rely on Facebook and Instagram business accounts for distribution and advertising. Their legal teams should review the October 30th terms closely, document their robots.txt and crawler-blocking settings, and record any objection in writing.
For policymakers, the episode strengthens the case for a statutory answer rather than leaving consent to platform contracts. The Department for Promotion of Industry and Internal Trade has been examining AI and copyright, and MeitY is shaping wider AI governance. Clear rules on whether training needs a licence would protect Indian publishers far more reliably than negotiating clause by clause with each global platform.
FAQ
When do Meta's new business terms take effect?
The revised terms for business accounts come into force on October 30th. Businesses that keep using Meta's tools after that date are treated as accepting them, so publishers that object should raise concerns and review their settings before then.
Did Meta remove the right to train AI on scraped content?
Meta removed the explicit word "train", but kept language allowing it to use content to develop business products. Legal experts including Matt Topic argue that "develop" is broad enough to include training. Meta says the change does not override robots.txt preferences, so the real-world effect remains contested.
What should Indian publishers do now?
Indian publishers with Meta business accounts should audit their robots.txt files and crawler-blocking rules, read the new terms with counsel, and keep a written record of any objection. Watching the ANI v OpenAI case in the Delhi High Court is also wise, as it may set the first Indian precedent on AI training and copyright.
Where can I read the original reporting?
The Verge published the original coverage, drawing on interviews first reported by Nieman Lab. The link appears in the attribution note below.
This story was reported by The Verge. Read the full original coverage at The Verge.