Twitch deploys real-time AI monitoring of livestreams, sparking dispute over ad payouts for “brand-unsafe” streamers

Twitch, the world’s largest livestreaming platform, has introduced a real-time AI monitoring system to identify content that does not suit advertisers and block the advertising revenue paid to the streamers behind it, drawing fierce backlash from creators.

CEO Dan Clancy says Twitch now uses AI to identify streamers that may not match a brand, since it now transcribes all streams for brand safety.

It won't show ads on channels or content the AI believes don't match what the advertiser is trying to achieve. pic.twitter.com/aGGq9fDb7U

— Zach Bussey 🇨🇦 (@zachbussey) September 29, 2026

■ CEO Dan Clancy reveals plan: “AI transcribes every word on streams to assess brand suitability”

Twitch CEO Dan Clancy recently appeared on The Snarketing Podcast and said the platform uses AI to transcribe audio from live broadcasts in real time. Twitch then uses that data to assess whether a streamer is providing content that is clean enough for brand advertising.

Twitch has traditionally operated content classification guidelines under which streamers applied their own tags when broadcasting games rated for audiences 18 and older or adult art. Clancy said that while creators’ self-labeling accuracy is as high as 99%, Twitch has recently made a major investment in using AI to directly determine and analyze whether a streamer aligns with an advertiser’s brand in an effort to ensure brand safety.

In practice, that means streamers who frequently use profanity, make political comments or tell sexual jokes that advertisers dislike could be caught by the AI monitoring system, excluded from ad placements or see their revenue cut sharply.

■ “We could broadcast freely without sponsors...” Streamers vent their anger

Creators on the platform strongly criticized Twitch after the news broke, accusing it of trying to exert excessive control over streamers’ everyday broadcasts.

SteveInSpawn, a Twitch partner streamer with about 30,000 followers, told entertainment publication Aftermath: “When I’m doing a sponsored stream, I avoid politics and keep things as tame as Sesame Street. But on my personal streams, I make funny jokes and let my rougher sense of humor out. That’s how every creator works.” He added: “We already distinguish for ourselves when we need to act in line with a brand and when we don’t. It’s absurd for AI to snitch on a sexual joke I made when I was by myself and take away my ad revenue.”

Concerns are particularly acute among LGBTQIA+ and virtual YouTuber, or VTuber, creators. LuminusRed, who has 16,000 followers, said, “Past experiments showed that ad revenue fell when we used queer-related tags such as ‘lesbian.’” LuminusRed added: “I’m afraid AI will ignore nuance, context and tone, hold up a blunt mirror and decide that a brand is offended, then cut our revenue.”

Streamers also say it is a major problem that Twitch may not transparently tell them why they were placed on a particular brand’s blacklist.

■ Twitch’s double standard after championing “raw authenticity”

In recent years, as AI-generated content has flooded the internet, Twitch has emphasized that the “raw authenticity” of creators onstage and real-time interaction are the platform’s greatest strengths.

But the new AI surveillance system is drawing criticism that it will destroy that raw appeal and force every creator to produce bland, odorless content that appears to have been vetted by a committee. Brand safety matters to advertisers, but growing numbers of critics say excessive AI censorship threatens the humor and autonomy at the heart of the livestreaming ecosystem.