The engine identifies distortions by evaluating grounded response segments against extracted source content to flag claims that the AI altered, misrepresented, or presented out of context, separating them from survived or lost facts.
flowchart LR
A[Source Webpage Content] --> C[Veracity Evaluation Prompt]
B[Grounded AI Response Segments] --> C
C --> D[Survived: Accurately Reflected]
C --> E[Lost: Omitted Relevant Facts]
C --> F[Distortions: Altered or Misrepresented]