Meta Pauses Controversial Terms That Could Expand AI Content Scraping

Introduction

Meta has indefinitely paused a planned update to its commercial terms that could have broadened the company’s ability to collect content from websites linked through Facebook business pages. Meta told Reuters that it was putting the changes on hold while continuing discussions with stakeholders. In the meantime, the current terms will remain in effect.

The decision temporarily removes an immediate pressure point for publishers and other business-page owners. It does not, however, settle the larger dispute over how platforms should obtain and use external content for artificial intelligence development.

Key points

• The proposed terms linked page access to external crawling. Business users were told that keeping a Facebook business page could require them to allow Meta to crawl websites linked from that page and use the resulting material to develop business products.
• The opt-out question was central. Legal experts warned that broad wording could be read as allowing Meta to disregard crawling preferences expressed through robots.txt. Publishers also objected to provisions understood to require the creation of test credentials that could let Meta’s bots reliably reach content behind paywalls.
• Publishers feared lost licensing leverage. Because news organizations commonly maintain Facebook pages, industry groups argued that the policy could give Meta access to large volumes of journalism without the kind of licensing negotiations publishers have sought from AI companies.
• The proposal drew international scrutiny. Courts in Brazil and Norway issued preliminary injunctions blocking the terms. The UK Competition and Markets Authority opened a formal investigation into whether Meta might be using a dominant position to obtain valuable AI training data. US lawmakers from both parties also urged Meta to withdraw the update.
• Meta says its crawling has not changed. The company has said that content owners can still communicate preferences through robots.txt. Critics responded that the revised terms did not contain specific language making that commitment clear or enforceable.

Why it matters

The pause is a policy retreat, but not a final resolution. Meta says it wants more consultation with industry groups and leaders before changing the rules. That leaves open the possibility of a later proposal that again seeks broader access to linked web content.

The dispute illustrates why simply replacing the word “train” with a broader term such as “develop” is unlikely to reassure publishers. The important questions concern the scope of collection, the permitted uses, treatment of paywalled material, and whether website operators have a practical way to refuse access. Without precise answers, a platform’s general assurances may not overcome concerns about compulsory data access.

The case also places AI data collection within a wider competition-policy debate. A platform that controls a major distribution channel can potentially make access to that channel conditional on permission to collect more data. Regulators and courts will therefore be examining not only copyright and licensing, but also whether commercial terms give a dominant platform unfair leverage over publishers and other businesses.

The next test will be Meta’s eventual replacement for the delayed terms, along with the outcome of the regulatory and court reviews already under way.

Source: Ars Technica AI