When I ask Grok about a developing news story, several X posts can look like stronger evidence than one. But those accounts might all be repeating the same screenshot or citing the same unnamed source.

I’d like to use it for source provenance checks, not just to count how many people are repeating a claim.

“Show which posts contain firsthand reporting, which cite another account, and which have no identifiable source.”

Is that a reasonable request, and how would you check the result? Separate wording does not necessarily mean separate reporting, while two eyewitnesses might describe the same event in very similar terms.

My main question is how to distinguish independent corroboration on X from a chain of reposted assertions. If the original source is deleted or inaccessible, can the response make that gap explicit rather than treating the remaining posts as confirmation?