A study by Martin Eling argues that artificial intelligence risks are harder to insure than standard cyber breaches due to internal system failures and widespread model dependency. The analysis frames AI risk as structurally different from conventional information security exposures, where failures can cascade across dependent components and where model behavior can create unpredictable outcomes. For healthcare organizations evaluating AI deployment, this may increase the need for insurance and risk transfer products explicitly designed for AI-specific failure modes. The near-term impact is practical: procurement, governance, and vendor risk reviews may increasingly require more explicit coverage for model-dependency and internal malfunction scenarios rather than only perimeter cyber threats.