Risk Management Show
Protecting AI Data: Key Risks and Solutions with Jason Rogers
Mar 11, 2025
Join Jason Rogers, CEO of Invary and expert in AI data security, as he shares his insights on safeguarding critical data from tampering and theft. Discover essential strategies like confidential computing and runtime attestation that can dramatically enhance data security. Jason emphasizes the importance of protecting training data, metadata, and inference data throughout their lifecycle. He urges organizations to reevaluate their data management practices and explore open-source initiatives to better guard their valuable information. Tune in for essential data protection tips!
11:46
Episode guests
AI Summary
AI Chapters
Episode notes
Podcast summary created with Snipd AI
Quick takeaways
- Securing AI data, especially inference data, is critical to prevent tampering and protect sensitive information that can impact AI outcomes.
- Confidential computing serves as an essential solution by encrypting data in use, thus safeguarding it during processing against potential threats.
Deep dives
Understanding AI Data Security
The discussion emphasizes the critical need for securing various types of data in the AI lifecycle, including raw training data, network weights, metadata, and inference data. Inference data, in particular, poses significant risks, as it may contain personally identifiable information (PII) and proprietary information. If this data is tampered with or stolen, it can undermine the accuracy of AI outcomes and lead to unwanted exposure of sensitive information. Organizations must treat this data with high priority and implement robust security measures to mitigate these risks.
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.
