Twitch has announced a new “clarification” to make its sexual harassment policy more understandable. This was announced in conjunction with an update to the AutoMod moderation tool designed to help streamers curb inappropriate messages in chat.

In a blog post published yesterday, Twitch wrote that its sexual harassment policy “remains largely unchanged,” but it has added clarifying language to better define the boundaries of the types of behavior that are not acceptable on its platform.

The new language is more specific about what Twitch does not want to see in chat. We prohibit unwanted comments, including comments about someone's appearance or body, sexual requests or advances, sexual objectification, negative remarks or attacks related to a person's perceived sexual behavior or activity, regardless of gender, and comments using emojis or memes. “"We also do not, under any circumstances, condone the recording or sharing of intimate images or videos without consent. ““Such content may be reported to law enforcement authorities.”” Read the updated guidelines.

In other words, Twitch is now attempting to provide viewers with a robust list of specific behaviors that can be used as a reference when deciding whether or not to post in chat, rather than a broader and more vague ban on creepy behavior.

The platform emphasizes that (as expected) it is the “non-consensual” elements that are really at issue and urges anyone who feels targeted by “mutually consensual comments” to object. However, he said that some comments that “express an intent to commit sexual violence” will never be allowed on Twitch, regardless of other factors.

Twitch has also updated AutoMod, a machine learning gizmo that streamers can use to automatically nip inappropriate comments in the bud, to “better combat sexual harassment.”

“We've developed a new AutoMod category that can filter out chat messages that could be considered sexual harassment.” Says Twitch. “This new category provides an additional layer of protection and helps block those messages on the fly before they appear in chat.”

Streamers can decide how severely AutoMod will withhold potentially problematic comments. Streamers, along with mods, can review the withheld messages and decide what to do with them. The new tool was released yesterday and is currently only available in English.

Of course, sexual harassment is absolutely rife in the online space, and Twitch is no exception. Numerous streamers (the majority of them women) have plenty of horror stories of viewers crossing the line in some way, whether it be creepy comments or outright stalking. Such tools are unfortunately necessary, but by no means sufficient to stop the problem altogether.