How it works

AI-assisted verification from claim to context.

TruthBubble is built around user-initiated checks, source-backed explanations, and confidence-aware verdicts for fast-moving digital claims.

1

User-initiated capture

TruthBubble starts when a user chooses content to verify, such as a post, screenshot, video, article link, or forwarded message. It is not positioned as continuous screen recording.

2

Claim identification

The app looks for factual claims that can be checked. Opinion, jokes, missing context, and ambiguous wording may need different treatment than direct factual statements.

3

Context reconstruction

A useful check often requires dates, source accounts, captions, linked articles, visible text, and surrounding context that may be missing from a viral repost.

4

Evidence retrieval

TruthBubble is designed to look for relevant evidence from sources, articles, public records, and corroborating material where available.

5

Visual support review

For screenshots and videos, visual evidence can support what appears in the content, but it is treated carefully because visibility is not always proof of the claim.

6

Confidence-aware verdict

The result should explain the verdict, confidence, sources, missing context, and limitations so users can decide what to inspect next.

Sources and limitations stay visible

TruthBubble should show the evidence and uncertainty behind a result. Learn more about the methodology and the limits of AI verification.