Twitch Uses AI to Decide Which Streamers Get Ad Money
Twitch has confirmed they are using artificial intelligence to help determine whether a streamer’s content is suitable for advertisers, raising concerns among creators whose income depends partly on ad revenue.
CEO Dan Clancy discussed the system during a recent appearance on the Snarketing Podcast, where he explained that Twitch transcribes what streamers say during broadcasts. Those transcripts can now help determine whether a stream fits an advertiser’s preferences and identify parts of a broadcast that a brand may not want its ads appearing alongside.
The system builds on content labels that streamers already apply to their broadcasts. According to Clancy, Twitch says creators use those labels in line with the platform’s classification guidelines with more than 99% consistency. Advertisers can then choose the types of content they are comfortable appearing next to.
The latest system arrives as Twitch is already facing questions about their growing use of AI. In August, the platform drew criticism after rolling out Amazon AI training, with users opting in by default. Twitch also uses AI or automated technology for features such as captions, stream summaries, generated clips, and safety tools.
For streamers, though, the bigger issue with ad matching is whether AI can understand context well enough to influence something as important as their earnings. Words that appear risky in a transcript may be harmless when heard as part of a joke, sarcastic comment, discussion, or conversation. Accents, slang, and other nuances could make those judgments even harder.
Some creators are already connecting the system to changes in their ad revenue. Longtime streamer JP “itmeJP” McDaniel suggested it could explain a recent decline in his earnings. VTuber LuminusRed said she had previously noticed lower ad revenue when using queer-related tags, raising concerns that discussions about identity could affect how advertisers view her content.
This change could heavily influence what streamers say and how they behave on air, potentially replacing some of the platform’s semi-authentic feel with content that is far more calculated around advertiser preferences. Twitch says AI will not have the final say over a creator’s long-term ability to earn ad revenue, though, with a company representative explaining that advertisers set their own preferences, meaning a channel may qualify for one campaign but not another.
The company also says staff review flagged content at multiple stages, with any long-term change to a creator’s ad eligibility ultimately decided by a person. Twitch is developing tools that will tell streamers when advertiser preferences are affecting their revenue, which could provide some clarity when earnings change.
For more articles like this, check out our Tech News page!