How AI forces us to expand our thinking about basic cybersecurity concepts: Part 2 – Confidentiality

The article discusses how to preserve confidentiality in AI/ML systems by aligning AI access rights with user permissions to prevent unintended disclosure, preparing and anonymizing training data to reduce privacy leakage and overfitting risks, and addressing intellectual property concerns related to patents, copyrights, and trade secrets (including techniques like code obfuscation). It concludes that organizations should adapt governance and security controls for AI in line with regulations and standards, balancing confidentiality with utility, and notes NVISO’s support for securing AI systems and achieving compliance.