AI agent re-grades 75% of high-severity cloud security findings

Sola Security tested an AI agent that re-checks cloud scanner findings against the actual environment. Across 9,967 HIGH findings from two commercial platforms in eight production environments, the agent changed the grade of 75.1%: 58.5% were downgraded and 16.6% upgraded.

• Agent changed the grade of 75.1% of 9,967 HIGH findings: 58.5% down, 16.6% up
• 599 vendor rules each issued one severity across 99,073 failing findings
• 32.3% of downgrades involved inert or orphaned resources
• Live AWS probes matched the agent's facts 99.4% of the time