Meta revises terms to drop disputed web‑scraping clause

Meta updated its commercial terms after feedback that a specific sentence suggested the firm could scrape content from news sites to train its AI models without regard to site‑level restrictions.

The removed clause read that Meta may use any content from listed domains to provide, develop, improve, research, train, and maintain its products. Critics, including legal scholars, argued this could override a website’s robots.txt directives and provide free training data.

Why it matters for GPU / AI infrastructure

Clear terms help AI teams plan data‑acquisition pipelines with confidence, reducing legal risk and unexpected changes in training‑data availability. For GPU‑heavy workloads, stable data sources mean more predictable compute utilization and better capacity planning for AI training clusters.