Meta Postpones New Business Terms Amid Publisher Backlash Over AI Scraping

Meta announced that the rollout of its revised business terms, originally set for October 30, will now take effect on November 20. The delay follows a coordinated pushback from news organizations, industry groups, and European regulators who argued the language could grant Meta broad rights to crawl and ingest publisher content for AI model training.

Although Meta removed explicit references to “training” from the terms, legal analysts said the remaining provisions still leave room for large‑scale scraping. The News/Media Alliance sent a formal letter to CEO Mark Zuckerberg demanding a rewrite, while EU and UK advertising bodies called on authorities to intervene.

Why it matters for GPU / AI infrastructure

Access to high‑quality, legally cleared text data is a primary bottleneck for training large language models on GPU clusters. Uncertainty over scraping rights can force AI teams to curate smaller, proprietary datasets, increasing compute costs and slowing model iteration. Clearer licensing frameworks reduce legal risk and help infrastructure providers like AiGpu plan capacity for compliant data pipelines.