On a livestream on the official Twitch channel, Head of Community Mary Kish and Chief Product Officer Mike Minton faced questions from nearly 3,000 viewers . When asked directly why the feature was opt-out rather than opt-in, Minton gave a remarkably blunt answer: "If it was opt-in, nobody would opt in"
.
Minton also argued that competitors like Google and Meta offer no opt-out choice at all, and noted that external scraping of public content would happen regardless . The response was widely covered by outlets including Forbes, which reported that Twitch effectively admitted they knew creators wouldn't voluntarily agree, and so chose a default that maximized data collection
.
Twitch executives failed to clearly clarify whether content had already been used by Amazon before the opt-out setting even existed. Multiple sources report that Amazon had been quietly using Twitch content to train its generative AI models for years — the August 2026 update simply gave streamers a visible toggle they could flip . TechSpot reported that Amazon had been doing this "years after it started" and that Twitch had "never been exactly forthcoming about this fact"
.
This left streamers uncertain whether opting out would retroactively prevent their already-scraped data from being used, or only stop future collection — a point Twitch never clearly clarified .
Both Twitch (owned by Amazon) and Meta (Facebook/Instagram) have faced major backlash for using user content to train AI models by default. Here's how they compare:
| Aspect | Twitch (Amazon) | Meta (Facebook/Instagram) |
|---|---|---|
| Default setting | Opt-in by default (must manually opt out) | Opt-in by default (must manually opt out) |
| User choice offered | Yes, an opt-out toggle exists | Yes, a formal objection/opt-out form exists |
| Retroactivity clarity | Unclear — past use was not acknowledged until this setting | Meta faced similar backlash and legal challenges in the EU |
| Key difference | Twitch's CPO explicitly admitted "nobody would opt in" — a rare level of candor | Meta generally frames its policy as part of "improving services" without similar blunt admissions |
A key distinction highlighted by Twitch executives during the live stream was that Meta and Google offer no opt-out choice at all for similar training, suggesting Twitch's approach was actually more permissive than its competitors — though that framing did little to calm the backlash . Meta, by contrast, faced its own major backlash in 2024–2025 for training AI on public Facebook and Instagram posts by default, particularly in the EU and UK where regulators intervened
. But Meta did not offer the same blunt admission that they knew users would say no
.
To stop Amazon from using your Twitch content for AI training, you must manually disable the setting. Here's how :
Note that opting out will not prevent your chat messages from being used if you participate in another streamer's channel who has not opted out — their channel setting controls third-party messages .