Agent Maturity Compass API - v1.2.0
    Preparing search index...

    Interface ProcessDeceptionDetectionReport

    Process Deception Detection Scoring Module

    Scores AMC-7.1 to AMC-7.12: whether an AI system tests for and resists alignment faking, sandbagging, scheming, and goal-directed deception.

    Part of the AI Safety Research Alignment evaluation lane.

    interface ProcessDeceptionDetectionReport {
        evidenceFound: string[];
        gaps: string[];
        level: number;
        questionScores: Record<string, number>;
        score: number;
    }
    Index
    evidenceFound: string[]

    Evidence artifacts found

    gaps: string[]

    Specific gaps and recommendations

    level: number

    Maturity level L0-L5

    questionScores: Record<string, number>

    Per-question scores (0-1) keyed by question ID

    score: number

    Overall score 0-100