Data Security Challenges
Eiman discusses the complexities of handling sensitive information when using language models, emphasizing that trust in model providers doesn't eliminate risks associated with the underlying compute platforms. He introduces the Stained Glass Engine, which transforms embedding vectors into probability distributions, enhancing data security by preventing the straightforward extraction of sensitive data in plain text. The conversation highlights the need for vigilance against common vulnerabilities like human error and phishing attacks.In this clip
From this podcast
Super Data Science: ML & AI Podcast with Jon Krohn
SDS 843: Safe, Fast and Efficient AI — with Protopia’s Dr. Eiman Ebrahimi
Related Questions