Bytedance algorithm training content scraping
Occurred: April 2022
TikTok parent company Bytedance scraped content from Instagram, Snapchat and other platforms and used them to train the algorithms of Flipagram, a TikTok predecessor, according to a Buzzfeed report.
Former employees and internal documents reveal Bytedance took hundreds of thousands of short-form videos, usernames, profile pictures, and profile descriptions without the knowledge of their owners or the social media platforms on which they had been posted.
Two employees told Buzzfeed the scraped content was later used to train the US version of TikTok's 'For You' recommendation algorithm.
Bytedance denies the allegations.
Purpose: Recommend content
Technology: Content recommendation system
Issue: Privacy; Copyright
Transparency: Governance; Privacy