Limitations

AI verification should be useful without pretending to be perfect.

TruthBubble is built to surface context and confidence, but automated analysis can still be incomplete, uncertain, or wrong.

Important limits

  • Relevant evidence may be unavailable, paywalled, deleted, newly published, or not indexed.
  • Sources can conflict, especially during breaking news or politically sensitive events.
  • Screenshots may omit dates, captions, source accounts, edits, or the original context.
  • Manipulated media, old footage, satire, and edited clips can be difficult to identify automatically.
  • Confidence is not certainty. It reflects available evidence and can change.
  • Results can improve or change when stronger evidence appears later.
  • Medical, legal, financial, safety, and emergency decisions require qualified expertise.
  • TruthBubble assists verification and does not replace professional fact-checkers or journalists.

Use results as a starting point

A TruthBubble verdict should help users decide what to inspect next. It should not be treated as a legal, medical, financial, or safety instruction. For methodology details, read how TruthBubble handles sources and confidence.