OpenAI and Anthropic Beef Up Customer Data Confidentiality for Frontier Models, but Details Lacking

Mikołaj Barczentewicz August 21, 2026

EU Tech Reg View Original Source

• 
  Back
  to Top

OpenAI and Anthropic beef up customer data confidentiality for frontier models, but details lacking

It looks like I’m not isolated in thinking a lot about ways in which AI services can be provided not only securely, but also confidentially—even vis-à-vis the service provider (see my AI privilege and ChatGPT encryption, Trustworthy privacy for AI: Apple’s and Meta’s TEEs, and Meta AI’s Incognito Chat: a new bar for confidential AI). One thing you may have noticed in my commentary on confidential AI solutions is that I didn’t discuss the two frontier AI labs.

Read the full piece here.