Meta's new terms could let it train AI on news sites for free
meta
| Source: The Verge | Original article
Meta's updated terms, effective Oct. 30, permit the company to crawl and scrape content from news sites linked to businesses using its platforms, enabling free AI training on that material.
Meta has revised its commercial terms for business accounts, a change that takes effect on 30 October. The updated language grants the company permission to crawl and scrape “associated” websites—explicitly including online publishers—so that the content can be used to “develop, improve, research, and train” Meta’s artificial‑intelligence products. The terms contain no opt‑out mechanism for businesses that link their sites to Meta’s platforms, meaning publishers could be compelled to supply their articles without licensing fees.
Legal analysts consulted by Nieman Lab say the wording could be interpreted as a blanket right for Meta to harvest news‑site content for AI training. If enforced, the clause would give Meta a vast, cost‑free data source that rivals the paid licensing deals other tech firms negotiate with media outlets. For publishers, the prospect of losing revenue from content licensing while seeing their work repurposed in proprietary models raises both financial and ethical concerns. The move also spotlights the broader debate over how much user‑generated and third‑party content platforms may exploit for AI development without explicit compensation.
What follows will likely shape the industry’s response. Media organisations are expected to issue statements or seek legal clarification, and some may explore collective action against the terms. Regulators in the EU and elsewhere may scrutinise whether the clause complies with emerging data‑use and copyright rules. Finally, observers will watch how Meta integrates the newly acquired text corpus into its AI products and whether the company’s approach prompts similar policy shifts among other platform operators.
Sources
Back to AIPULSEN