TikTok, an app that was erstwhile conscionable a spot for user-generated content, is launching a caller mounting that lets users take really overmuch AI-generated contented they want to spot successful their “For You” feed. The institution is besides introducing much precocious labeling technologies for AI-generated content.
The caller AI-generated contented (AIGC) power is rolling retired wrong nan app’s “Manage Topics” tool, which lets users take what they spot connected TikTok.
“Manage Topics already enables group to set really often they spot contented related to complete 10 categories for illustration Dance, Sports, and Food & Drinks,” TikTok explained successful a blog post. “Like those controls, nan AIGC mounting is intended to thief group tailor nan divers scope of contented successful their feed, alternatively than removing aliases replacing contented successful feeds entirely.”
The move comes arsenic companies for illustration OpenAI and Meta are embracing AI-only feeds. In September, Meta released Vibes, a caller feed for sharing and creating short AI-generated videos. A fewer days aft Meta’s launch, OpenAI released Sora, a societal media level for creating and sharing AI-generated videos.
Since Sora’s launch, realistic AI-generated videos person been posted to TikTok. Additionally, galore TikTok users are leveraging AI to create visuals for posts astir different topics, for illustration history aliases celebrities.
TikTok says that pinch nan caller AI-generated contented control, users who want to spot little of this benignant of contented tin now dial things down, while those who bask it tin take to spot much of it.
Image Credits:TikTokYou tin entree nan caller capacity by going into your Settings, selecting “Content Preferences” and past clicking nan “Manage Topics” option. Then, you tin move nan slider for different topics, including AI-generated content, to set really overmuch you do aliases don’t want to spot that benignant of contented successful your For You feed.
The alteration is rolling retired successful nan coming weeks, TikTok says.
To amended its expertise to explanation AI-generated content, TikTok is now testing a exertion called “invisible watermarking.”
TikTok already requires group to explanation realistic AI-generated contented and uses a cross-industry exertion called Content Credentials from C2PA, which embeds metadata into contented that lets it and different platforms cognize erstwhile thing is AI-generated. However, TikTok notes that these labels tin beryllium removed erstwhile contented is reuploaded aliases edited connected different platforms.
With nan caller “invisible watermarks,” TikTok will adhd different furniture of safeguards by utilizing a watermark that only it tin read. That intends it’ll beryllium harder for others to region it.
TikTok will commencement adding invisible watermarks to AI-generated contented made pinch TikTok devices for illustration AI Editor Pro. It’s besides adding them to contented uploaded pinch C2PA’s Content Credentials. The institution says these watermarks will thief it explanation contented much reliably. TikTok notes that it will proceed reference C2PA’s Content Credentials and adhd them to AI-generated contented made connected its platform.
Related to these efforts, TikTok besides announced that it’s launching a $2 cardinal AI literacy money aimed astatine experts, for illustration nan nonprofit Girls Who Code, to create contented that teaches group astir AI literacy and safety.
1 day ago
English (US) ·
Indonesian (ID) ·