Being yourself on Twitch is a little more complicated now, and that has some creators worried about their ad dollars.
Twitch CEO Dan Clancy says the platform is using AI to analyze livestream transcripts and help determine whether a creator’s content meets advertisers’ brand-safety standards. That could have a bigger impact on creators than simply changing how Twitch moderates content.
Why? Because brand safety and image can affect revenue. Advertisers don’t want their ads appearing next to content they consider risky or inappropriate.
Creators are understandably concerned about what the AI system considers “brand safe,” and what happens when an algorithm gets the context wrong. Context matters because livestreaming is supposed to be, well, live. Creators swear. They make weird jokes, talk about politics, get emotional and go off-script.
That unpredictability is part of the appeal.
Is this brand safety, or censorship by algorithm?
Some streamers have already raised concerns about everything from LGBTQIA+ tags to the difficulty of teaching AI to understand context. Can the program detect sarcasm, inside jokes and the reasoning behind something a streamer says? What happens when a conversation about politics or social issues gets interpreted differently by an algorithm than it would by a human?
More importantly, will creators get to know why their content isn’t considered a good fit for certain advertisers?
These are not small questions. Twitch doesn’t exactly have a great track record of being transparent about its decisions and processes.
Creators who rely on sponsorships and ad revenue are facing an uncomfortable balancing act. How much of your personality should you sanitize for the algorithm? If your audience loses interest and stops watching because of it, was it even worth sanitizing?
Twitch hasn’t said that creators need to change how they behave on stream. As AI takes a bigger role in deciding where advertising dollars go, creators may want to think carefully about where they stand on this one.