Scanners#

Quadrant IntegrityLens uses a collection of rule-based scanners to detect patterns characteristic of AI-generated text. Each scanner focuses on a specific type of indicator.

Scanners fall into five categories:

Confidence levels#

Each finding has a confidence level:

LevelMeaning
HighStrong indicator of AI-generated text
MediumLikely AI-generated, but could occur naturally
LowWeak indicator — consider alongside other findings

A single Low-confidence finding means little on its own. Multiple findings across different scanners paint a clearer picture.

Disabling scanners#

Any scanner can be disabled with the --disable flag using either the scanner ID (disables the entire scanner) or a rule ID (filters individual findings):

integritylens --disable swiss-german document.pdf
integritylens --disable eszett document.pdf