xAI has added custom moderation controls for text API requests to its Chat Completions and Responses APIs. Developers can include a `guidelines` array to tell Grok how to handle content in three categories: NSFW content, violence and profanity.
xAI says Grok follows those guidelines while writing its response. The documentation does not specify how the controls behave in edge cases or what guideline values are accepted.
Why it matters
Developers using these APIs can tailor Grok’s handling of those content categories, though the documentation does not detail the limits of that control.
