Meta has removed language from its terms of service that sparked concerns about the company's web scraping and AI model training practices. The change came after the revised terms generated criticism over passages that appeared to expand Meta's ability to collect data from the internet for artificial intelligence development.
A Meta spokesperson stated that the updated terms did not reflect a change in how Meta actually crawls the web. The removal suggests Meta responded to public pressure regarding transparency about its data collection methods, though the spokesperson's statement indicates the practical scraping operations remain unchanged.
The specific language that triggered concern involved provisions about how Meta could use publicly available information. Critics worried the phrasing gave Meta broader latitude to harvest data from across the internet without explicit user consent, potentially feeding into training datasets for AI systems.
Meta's decision to revise the language represents an effort to clarify its policies rather than fundamentally alter its operations. By removing the contentious passages, the company attempted to reduce confusion about its data practices while maintaining its existing scraping infrastructure.
The incident highlights ongoing tensions between tech companies' data collection needs for AI advancement and public concern about privacy and transparency. Meta joins other major technology firms facing scrutiny over how they acquire training data for large language models and other AI applications.
The spokesperson's clarification that operations remain unchanged suggests Meta views the revision primarily as a communication adjustment. However, the removal of specific language may provide clearer guidance to users and regulators about what Meta does and does not do with web-scraped data.