Glossary term
Glossary term
Safety and Alignment
Risk of large-scale harm that can spread across users, sectors, markets, infrastructure, or society. In AI governance, systemic risk is especially relevant to GPAI, frontier models, mass deployment, and highly autonomous systems. Systemic risk analysis considers scale, interconnectedness, dependency, misuse potential, concentration of providers, and impact on critical services.
EU AI Act Article 55 imposes additional obligations on GPAI models with systemic risk including model evaluation, adversarial testing, and incident reporting.
The IMF and Bank of England have published 2024 papers on systemic financial risk from AI concentration among a few model providers.
The CrowdStrike outage of July 2024, while not AI-specific, illustrates the systemic risk pattern of dependence on concentrated software providers.