Validating GenAI Models – From Hype to Control

Key takeaways

This webinar covered:

• Why Generative AI introduces new risks that traditional model validation methods cannot fully address, requiring a broader socio-technical governance approach.
• Why validating GenAI systems extends beyond statistical accuracy to include reliability, safety, explainability, and trustworthiness.
• How regulatory frameworks across regions share core principles, including governance, transparency, and risk-based proportionality.
• Why the complexity of GenAI components, including foundation models, fine-tuning, and runtime parameters, requires layered, application-specific validation strategies.
• How AI agents with memory and decision-making capabilities require validation of long-term, multi-step interactions and potential aging effects.
• Why the industry is shifting from experimentation to responsible governance through continuous monitoring, transparency, and management of shadow models outside formal controls.
   

Missed the webinar? Access the replay here.

Missed the live session? Watch the webinar replay: Crisil Integral IQ webinar on Model Risk Management: Operationalizing E-23: Practical Frameworks for Banks and Insurers

April 22, 2026