Quick Answer
Twitch’s new Training for Generative AI setting is enabled by default, allowing Amazon to use eligible channel material for future generative AI model training unless you opt out. Disable the setting under Settings > Security and Privacy if you do not want streams, clips, VODs, chats, and channel text or pictures included. The change affects every account, so check your settings directly.
Key Takeaways
- Twitch added the Training for Generative AI toggle on August 12, 2026.
- The toggle is enabled by default for every Twitch account.
- The setting covers streams, VODs, clips, chat, pictures, and text on a channel.
- Twitch says the opt-out concerns Amazon model training, not Twitch training its own generative AI models.
- Creators must manually open Security and Privacy settings to turn the option off.
What changed with Twitch AI training?
Twitch added a new account setting called Training for Generative AI on August 12, 2026. The setting permits Amazon to use specified Twitch channel material in future training for a model that generates or synthesizes text, audio, images, or video, unless the account holder disables the option.
Twitch made the setting available under Settings > Security and Privacy, but the company enabled it by default rather than asking creators to grant permission first. Multiple reports published on August 12 and August 13 described the setting as applying across Twitch accounts, which means inactive accounts and small channels need the same manual review as large creator accounts.
Twitch Chief Product Officer Mike Minton addressed the default-on design during a Patch Notes livestream, saying that an opt-in approach would not attract participation. Twitch Head of Community Mary Kish also told viewers that the company did not expect them to be happy or excited about the decision. Those comments matter because the controversy concerns both the scope of AI training and the lack of an affirmative consent step.
The practical response is to inspect the setting even if you do not regularly stream. Creators who have uploaded clips, maintained VODs, or used channel chat still have content associated with their accounts.
What Twitch content can Amazon use for generative AI training?
Twitch says the generative AI training option can cover streams, VODs, clips, stream chats, and pictures and text on your channel. The company’s wording says that this material may be used in future training for an Amazon-developed model, which means the notice describes potential use rather than confirming that every eligible item has already entered a training dataset.
The broad list matters because Twitch content contains more than recorded gameplay or livestream video. A channel can include spoken commentary, viewer messages, usernames, visual overlays, uploaded images, titles, descriptions, and community discussions. A creator who focuses only on stream footage could miss other material that remains visible through the account.
| Content category | Included in Twitch’s description | Practical privacy consideration |
|---|---|---|
| Livestreams | Yes | Streams can contain voice, video, gameplay, and on-screen material. |
| VODs and clips | Yes | Older recordings can remain associated with a channel after a live broadcast ends. |
| Stream chats | Yes | Chat can include viewer conversations and other user-generated text. |
| Pictures and channel text | Yes | Channel descriptions, images, and related presentation material can reveal creator identity or brand information. |
Twitch’s language does not provide a public list of individual Amazon models, a training schedule, or a method for creators to identify which specific items may be selected. The most sensible approach is to treat the toggle as a choice about future eligibility, not as a detailed audit of every piece of content that Amazon may process.
How do you opt out of Twitch AI training?
To opt out of Twitch AI training, open your Twitch account settings, select Security and Privacy, find Training for Generative AI, and switch the setting off. Twitch places the control inside the account privacy area, and the company’s Security and Privacy settings page is the direct location to check while signed in.
- Sign in to the Twitch account that owns or manages the channel.
- Open your account settings from the profile menu.
- Select Security and Privacy.
- Locate the Training for Generative AI setting.
- Turn the setting off and confirm that the control remains disabled.
Twitch AI training requires a separate check for each account because account-level preferences do not necessarily transfer between personal, brand, moderation, or alternate streaming accounts. Review every account that has posted content or has access to a channel, especially when a team shares streaming responsibilities.
Take a screenshot after changing the setting for your own records. Twitch has not publicly described an account history page that shows when the preference changed, so a dated screenshot can help document the choice if the setting later changes location or wording.
What does turning off the Twitch AI training setting do?
Turning off the Twitch AI training setting tells Twitch that you do not want the listed channel material used for Amazon’s future generative AI model training. The setting is an opt-out from the program described in Twitch’s notice, rather than a deletion request for all content already hosted on Twitch.
Twitch’s wording is important because it refers to future training. The available information does not establish whether material was collected before a creator changed the setting, whether previously processed material can be removed from datasets, or whether Amazon has already used any particular stream, clip, chat message, picture, or text item. Creators should not assume that disabling the option erases past uploads or removes content from every Amazon system.
Twitch also continues to need data processing for ordinary platform functions, such as delivering video, moderating content, operating chat, and enforcing platform rules. The opt-out does not mean a channel becomes private, and it does not replace normal account-security practices. Creators who discuss AI disclosures on other platforms may have seen similar concerns around AI content labels, where clearer notices can affect how audiences understand creator material.
The practical limit is straightforward: disable the training setting if you do not want future eligibility, then separately manage VOD visibility, clips, chat moderation, and account access according to your privacy needs.
Is Twitch training its own generative AI models on streamer content?
Twitch says Twitch is not training generative AI models on streamer content, but the company says the new setting allows users to opt out of Amazon’s generative AI content model training. The distinction comes from Twitch’s official support account, which described the setting as an Amazon program rather than a Twitch-operated model-training effort.
The distinction matters because Amazon owns Twitch, yet the two entities can operate different products, teams, and AI development programs. A streamer evaluating the setting should focus on the practical result: Twitch account material may be available for future training of an Amazon-developed generative AI model when the account preference remains enabled.
Twitch’s official support account published the clarification on August 12. The clarification does not narrow the listed content categories or change the default-on status, so creators who object to Amazon training use still need to disable the setting themselves.
Amazon’s AI activity also has broader consumer consequences beyond Twitch. Infrastructure demand from large AI programs can affect costs and public policy, including concerns explored in TechJournal’s report on AI data center electricity costs. The Twitch setting is narrower, but it shows how consumer-created data can become part of a larger AI supply chain.
Why are Twitch creators concerned about the default-on choice?
Twitch creators are concerned because the Training for Generative AI option starts enabled, placing the responsibility on users to discover and reject the program. Streaming content can contain a creator’s voice, visual style, community interactions, and creative work, so the decision has privacy, labor, and consent implications even when a channel does not show sensitive personal information.
Content creator and streaming reporter Zach Bussey first publicized the buried setting with a screenshot on August 12, 2026. The screenshot spread widely on social media because creators objected to the setting’s location and default state, rather than only to the existence of an AI training option.
The default setting also creates a practical issue for casual users. A person who streamed occasionally, posted clips years ago, or uses Twitch primarily for chat may not read platform announcements or routinely review privacy controls. A system that depends on manual opt-out can therefore include users who never made an informed, affirmative choice.
Creators who rely on a recognizable voice, visual format, or community should consider whether generative AI training fits their own boundaries. The answer will differ by creator, but the safest privacy choice is to turn the setting off when the creator does not want Amazon to use the listed materials for future model training.
What should Twitch streamers do after opting out?
Twitch streamers should confirm the AI training preference, review who can access their account, and check what content remains public. The generative AI toggle addresses one stated training use, but account security and content visibility still determine who can view, copy, or misuse material available on a channel.
Review VOD and clip settings after opting out, especially if old broadcasts include personal details, private conversations, or material that no longer reflects your current channel. Delete or restrict content you no longer want publicly available. Twitch’s privacy notice provides broader information about how the platform handles personal information, although it does not replace the specific AI training preference.
Enable strong account protection, including a unique password and available two-factor authentication options. Account compromise creates a separate risk because an attacker could alter channel settings, publish content, or access private messages. Users who install browser tools for streaming management should also be careful, because fraudulent extensions can redirect or monitor browsing traffic, as shown by reports about fake Chrome extensions.
Contact Twitch Support if the Training for Generative AI control is missing, will not save, or appears to turn back on after you disable it. Do not share account passwords, authentication codes, or recovery codes with anyone claiming to fix a Twitch setting through chat or email.
Does opting out stop viewers from copying Twitch content?
Opting out of Twitch AI training does not stop viewers or third parties from copying content that remains publicly accessible. The setting concerns Amazon’s stated future generative AI training use, while public streams, clips, screenshots, and chat messages can still be recorded or redistributed by viewers under separate platform rules and legal considerations.
Twitch creators who need stronger control should combine the opt-out with audience and content-management choices. Limit VOD availability, moderate chat, remove sensitive clips, avoid displaying private information on stream, and consider whether a public broadcast is appropriate for material that cannot be safely shared once recorded.
Creators should also recognize the limit of platform controls. A privacy setting can reduce one authorized data use, but it cannot guarantee that every viewer respects creator boundaries. Report impersonation, stolen clips, or account abuse through Twitch’s reporting tools, and seek legal advice for serious copyright, harassment, or identity-related disputes.
FAQ
Is Twitch AI training enabled by default?
Yes, Twitch enabled the Training for Generative AI setting by default when it added the control on August 12, 2026. Creators must manually turn the option off under Settings > Security and Privacy.
Where is the Twitch AI training opt-out setting?
The Twitch AI training opt-out setting appears under Settings > Security and Privacy in a signed-in Twitch account. Look for the control labeled Training for Generative AI and switch it off.
What content does the Twitch AI setting cover?
The Twitch AI setting covers streams, VODs, clips, stream chats, and pictures and text on a channel, according to Twitch’s stated language. The notice says those materials may be used in future training for an Amazon-developed generative AI model.
Does opting out delete old Twitch streams?
No, opting out does not delete old Twitch streams, clips, or VODs from your channel. The setting concerns future generative AI training eligibility, so remove or restrict older content separately if you no longer want it publicly available.
Is Twitch using creator content to train its own AI?
No, Twitch says it is not training its own generative AI models on streamer content. Twitch says the toggle concerns Amazon’s generative AI content model training, which is why creators who object should disable the default-on setting.