I am fairly new here...I keep having debates with my colleagues over how to prevent sensitive data from flowing into AI. I would consider data anonymization beats all other methods of preserving PII when working with AI. Are there better and more effective solutions to this? What would you suggest if one had to build their own AI model?

Would appreciate all POVs on this...many thanks!