Facebook's new Creator Studio app — rolled out to iOS creators in the US and Canada on August 12, 2026, after a smaller test that began in June — includes an AI tool that drafts replies to comments in a creator's own tone. The creator still has to read, edit, and approve each draft before it posts, so nothing goes out on autopilot. What's automatic now is the drafting step itself: a creator working through a comment section gets a ready-made reply sitting in front of them for almost everything, including comments under a paid partnership post. That's the part worth building into a brief now — not because the tool is unsafe, but because "the creator approved it" stops meaning "the creator thought carefully about it" once approving takes three seconds instead of three minutes.
What exactly did Facebook launch, and does it touch sponsored posts?
The app has two AI pieces. A creator assistant answers questions like "when should I post?" and "what are people saying in my comments?" based on the creator's own content style, performance, and audience — a research tool, not a public-facing one. The comment tool is the one that matters for brand work: it surfaces the comments Facebook judges most important and writes a reply in the creator's tone, which the creator can edit or accept before it's posted under their name. Facebook's own framing treats this as a general creator productivity feature, not something scoped to organic content only — nothing in the rollout distinguishes a comment under a regular post from a comment under a disclosed paid partnership. If a creator you work with has the app, the draft-and-approve flow is sitting under your campaign's comments the same as under everything else they post.
Which comments should never go through draft-and-approve, even with human review?
Human review catches typos and off-key tone. It doesn't catch a reply that's factually wrong about something the creator was never briefed on in detail, because the AI is drafting in the creator's voice, not the brand's current price sheet. Four categories are worth naming explicitly as no-draft, brand-only:
- Disclosure questions. "Is this an ad?" needs the exact wording your legal team already cleared, not a tone-matched paraphrase a creator approves in passing.
- Price, stock, and ingredient questions. These are facts that change on the brand's schedule, not the creator's — an AI draft trained on past comments has no way to know this week's price moved or a size sold out an hour ago.
- Refunds, complaints, and anything that reads like a legal or safety concern. These need a person who can actually resolve them, not a reply that sounds sympathetic and resolves nothing.
- Any follow-up on a claim from the brief the creator didn't originate. If a comment pushes back on a claim you gave the creator to say, the AI draft is guessing at your intent from one post's context, not the brief.
What should the brief or contract say about comment handling now?
Two additions cover most of the exposure. First, list the categories above by name in the brief and say plainly that they get routed to the brand's contact before any reply goes out, drafted by AI or typed from scratch — this is the same logic as putting disclosure wording in the contract rather than a guideline, because a routing rule only holds up if it's a term, not a suggestion. Second, give the creator a fixed phrase to use verbatim for the one or two questions you can predict — usually the disclosure one — so there's a lower-effort correct answer sitting right next to the AI's draft, not just a request to "review carefully" that competes with a tool built to make reviewing feel unnecessary.
Example: a creator running one sponsored post might see 200 comments over the campaign's first week. At even 20 seconds of real consideration per AI-drafted reply, that's over an hour of comment work — long enough that the ones needing more scrutiny quietly get the same 20 seconds as the ones that don't. Naming the four categories above turns that into a much smaller sorting job: most comments still get the fast AI-assisted pass, and only the handful that match a named category get pulled out for a slower, brand-checked answer.
How would a brand even know a creator is using this on a live campaign?
You wouldn't, directly. The tool lives inside the creator's own app, the same way TikTok Shop's AI-voice restrictions or Twitch's AI-training toggle sit entirely on the creator's or platform's side of the relationship — there's no account-level setting a brand can check from outside. The practical workaround isn't detection, it's disclosure: ask, as a standard campaign-kickoff question, whether the creator uses AI-assisted reply tools, and put the routing categories in the brief regardless of the answer, since the same categories deserve brand-only handling whether the reply was AI-drafted or hand-typed. Watching what's actually live under each post — not just what was originally posted — is the other half of it; a brand that only reviews the sponsored post itself at launch has no way to catch a wrong reply sitting in the comments three days later.
What if an AI-drafted reply already went out wrong?
Treat it the way you'd treat any wrong reply from a creator, AI-assisted or not: correct it in public rather than deleting it. A factually wrong reply about price or ingredients that quietly disappears reads, to anyone who saw it, like the brand had something to hide — a follow-up comment with the correct information, posted by the creator or the brand depending on who's better positioned to answer, closes the loop without the "what did they delete" problem a deleted comment tends to create. Then feed it back into the brief: whatever comment type produced the wrong reply moves into the named, brand-only list for next time, the same way any other recurring trigger would.
The tool itself isn't the risk — a human still has to approve every reply before it posts. The risk is that "approve" gets faster and quieter than "write," and a brief that doesn't name what's off-limits is trusting that speed won't matter. Want to see what's actually live under every sponsored post your creators have running, comments included? Get started.
