Home / Technology / Meta AI Training Data Row: Terms Softened, But Concerns Linger
Meta AI Training Data Row: Terms Softened, But Concerns Linger
23 Oct
Summary
- Meta adjusted business terms to remove direct AI training language.
- Legal experts remain unconvinced by Meta's revised terms.
- Meta states robots.txt preferences will be respected by its systems.
Meta has updated its business terms, removing previously controversial language that suggested it could freely train its AI models on content from associated news sites. This move comes after significant backlash from the publishing industry.
The initial terms, effective October 30th, stated Meta could use any content from business-associated websites to 'develop, research, and train' its products. Legal experts warned this could override existing blocking instructions like robots.txt.
While Meta has since removed the specific mention of 'training' its products with scraped content, some legal analysts believe the revised wording may not fully resolve the issue. Broad terms like 'develop' could still be interpreted to include AI training.
Meta maintains its updated terms are not intended to scrape news content without permission. A company spokesperson reiterated that robots.txt preferences are respected and that the updates do not supersede these signals.
