Twitch's harassment policy protects the person, not the face

The setup

On 9 December 2020, Twitch announced an updated Hateful Conduct and Harassment Policy, stating the new rules would take effect on 22 January 2021 after a stated education period during which the previous rules stayed in force. The company restated the policy's continued force in a 1 November 2024 post, Hate and harassment have no place on Twitch, calling the Community Guidelines foundational and confirming enforcement continues under them. Both documents remain accessible as retrieved on 16 September 2026.

What the documents show

The 2020 announcement defines hateful conduct as harassment that attacks a protected characteristic, and lists race, ethnicity, national origin, religion, sex, gender, gender identity, sexual orientation, disability and several other identity attributes as protected. It states explicitly that making lewd or explicit comments about anyone's sexuality or physical appearance is prohibited and that Twitch does not make an exception for public figures, meaning a streamer's public role does not reduce the protection. Neither the 2020 post nor the 2024 restatement mentions avatars, VTubers or on-screen presentation at all; the protection is defined by the identity characteristics of the person behind an account, not by what that account visually shows on stream.

The craft

For a VTuber who streams behind a Live2D or 3D model, this matters in a specific, narrow way the documents support directly: a comment attacking the performer's presumed gender, ethnicity or disability status is covered by the same protected-characteristic language whether the channel shows a webcam feed or an animated character, because the policy attaches to the person, not the pixels. What is not addressed by either document is how Twitch's moderation systems detect or verify identity-based harassment aimed at an avatar-fronted account in practice; the text states scope and protected categories, not enforcement mechanics for this specific case.

Keeping the creator in control

A performer weighing whether to report targeted harassment has a documented basis to do so regardless of how their channel presents visually, which is a meaningful floor of protection to know exists before an incident happens rather than during one. This is an editorial reading beyond the text: because the policy is characteristic-based rather than presentation-based, a creator does not forfeit its protection by choosing to perform through a character, and reporting a violation does not require disclosing anything about the person behind the avatar beyond describing the attack itself.

• Does the harassment target a protected characteristic of the performer, as Twitch defines that term, rather than general criticism of content?
• Has the incident been reported through Twitch's normal channels regardless of whether the account streams as a face or an avatar?
• Would confirming the report require disclosing information about the performer's real identity beyond what the attack itself already referenced?

Twitch's own documents draw the protection around the person, and an avatar-fronted channel does not sit outside that boundary merely because it is harder, from the outside, to tell who is behind it.