Methodology
How Toxispy detects rhetorical manipulation.
Detection Model
Toxispy uses language models to analyze text for rhetorical manipulation patterns. Each detected technique is classified by type and rated for severity.
Scoring System
The manipulation score is calculated as the number of detected tricks multiplied by their individual severity, relative to the text length.
Severity Levels
Low — decorative rhetoric, not harmful
Medium — persuasion techniques that distort framing
High — manipulative patterns designed to override critical thinking
Limitations
No automated system is perfect. Toxispy may miss subtle manipulation or flag legitimate rhetorical techniques. The tool is designed to assist human judgment, not replace it.