A source-backed AI agent that can receive a claim in context, assemble relevant evidence, and return a reviewable record without presenting automation as final authority.
The agent reduces the distance between a questionable claim and the evidence needed to examine it. It is designed for conversation surfaces and monitored workflows, including X, where a mention can become a structured fact-checking request.
Live demo: experience the agent on X via @0xProvenance. X is one live implementation; the agent can also operate across other platforms and defined organizational workflows.
The agent identifies the checkable question, searches for relevant sources, and produces a linked record that keeps evidence and reasoning available beyond the original exchange.
It supports human judgment rather than replacing it. Developing events, limited sources, conflicting accounts, and uncertainty remain visible instead of being compressed into false confidence.
Bring a claim from the conversation rather than forcing people into a separate research workflow before they can ask whether it holds up.
The response is built around relevant sources and a review record, not an unsupported answer detached from where its evidence came from.
Automation can accelerate collection and synthesis. Human review, domain expertise, and accountable policies remain essential where the stakes require them.
The same agent pattern can support defined streams, dashboards, APIs, and review queues for organizations monitoring high-volume information.
On 18 January 2026, IndiaTimes published a fact-check about a viral claim involving an alleged early GTA 6 access grant. Its article embeds a public assessment from Provenance Fact-Check. The citation documents this public output, not an endorsement. This case study describes the evidence reported at that time.
The circulating claim followed a family member’s LinkedIn appeal. The article describes social posts and reporting that referred to a now-deleted LinkedIn post; it does not establish that the alleged grant occurred.
The article reports that the claim circulated on X and other social channels, notes references to a LinkedIn post that had been deleted, and says there was no official Rockstar confirmation at publication.
Read the IndiaTimes articleThe article quotes the public post’s verdict as “Unverifiable.” The excerpt mentions X posts and a YouTube video, but is truncated; a complete source list or execution record is not available in the cited excerpt.
Inspect the original postThe public material cited here does not establish whether the alleged access was granted. No official confirmation was reported in the article at publication. “Unverifiable” is not proof the claim was false, and this historical snapshot does not establish the claim’s present-day status or prove system-wide accuracy.
Tell us about the conversation surface, review policy, and evidence requirements that matter to your organization.
Discuss the AI Agent