Under an adaptive attacker performing three iterations of targeted adversarial skill refinement against the final guard, SkillSonar reduces overall ASR from 45.1% to 22.9% (-22.2 percentage points), with reductions across all six individual risk families.

Under an adaptive attacker performing three iterations of targeted adversarial skill refinement against the final guard, SkillSonar reduces overall ASR from 45.1% to 22.9% (-22.2 percentage points), with reductions across all six individual risk families. · CiteArk

Measurement results

0/21 assessed

0 supported0 disputed or conflicting21 not assessed

The complete result group stays with its claim. Assessed rows come first; their judgments do not replace the whole-claim conclusion. Expand a row for conditions, sources, and evidence history.

Each row is a measurement with its reported value, latest available observation, and cumulative evidence state.
Metric and conditions	Reported	Latest observation	Evidence judgment
defense: No Guard · risk_family: Capability Control	55 percentage_points	—	Not assessed
defense: SkillSonar · risk_family: Capability Control	26.7 percentage_points	—	Not assessed
risk_family: Capability Control	-28.3 percentage_points	—	Not assessed
defense: No Guard · risk_family: Specification Integrity	53.3 percentage_points	—	Not assessed
defense: SkillSonar · risk_family: Specification Integrity	43.3 percentage_points	—	Not assessed
risk_family: Specification Integrity	-10 percentage_points	—	Not assessed
defense: No Guard · risk_family: Resource & Reliability	45 percentage_points	—	Not assessed
defense: SkillSonar · risk_family: Resource & Reliability	26.7 percentage_points	—	Not assessed
risk_family: Resource & Reliability	-18.3 percentage_points	—	Not assessed
defense: No Guard · risk_family: Privacy & Data Flow	43.3 percentage_points	—	Not assessed
defense: SkillSonar · risk_family: Privacy & Data Flow	8.3 percentage_points	—	Not assessed
risk_family: Privacy & Data Flow	-35 percentage_points	—	Not assessed

Under an adaptive attacker performing three iterations of targeted adversarial skill refinement against the final guard, SkillSonar reduces overall ASR from 45.1% to 22.9% (-22.2 percentage points), with reductions across all six individual risk families.

Reported and observed measurements

Assessments (0)

Runs (0)

Measurement results