Three Frameworks Shield CISOs From AI Liability
CISOs increasingly face liability questions from regulators, litigators, and customers when AI systems fail. Daniel B. Garrie, CEO of Law and Forensics, developed three frameworks to help security leaders assess whether specific AI deployments are safe and create defensible documentation of their risk decisions. The work builds on JAMS AI Rules, established in June 2024 to guide arbitration of AI disputes.
The first framework scores AI deployment risk across five dimensions: potential harm, data and rights, autonomy, evidence gap, and vendor dependence. Scores place tools into four risk categories—standard review, conditional approval, restricted pilot, or hold release. The second framework requires granular board documentation including scope, timeframe, risk exposure, deployment conditions, accountable parties, challenges raised during decision-making, and tool ownership. The third framework creates an incident response checklist to preserve critical evidence: system state at the time of the incident, interaction records, data and access information, governance records, and chain of custody details.
Garrie emphasized securing evidence preservation rights in vendor contracts, including access to model snapshots and clear risk allocation terms. He challenged CISOs to test the frameworks on one deployment, improve one board report, and run one tabletop exercise within 30 days. Liability ultimately hinges on what organizations can prove they did to identify and mitigate AI risks.
Read the full article
Continue reading at Dark Reading →This is an AI-generated summary. Read the original for the full story.