How do you stop users reading an AI’s stock inferences as a creator’s picks?

Most finance TikToks never name a stock. “The grid is the next AI trade” and that’s it.

So the obvious AI feature is: take the video, show what the creator actually said (quoted from the subtitles), then show the companies a model infers from it, each labelled as an inference with a source and a case against.

The problem: even with the label, people screenshot the second block and call it “his picks”. That’s a creator-attribution mess waiting to happen.

Which design actually fixes it?

A) Hide inferences until the user clicks to reveal

B) Show them only pinned next to the quote they came from

C) Make the user type their own belief first, then infer from that

If you’ve shipped AI output in a regulated-ish flow: A, B or C, and what label actually worked for you?