How to Use AI With Your Privacy Intact
wired ·
The article highlights the significant privacy risks associated with mainstream AI chatbots, which often collect and store highly personal user data for training or legal compliance. Cryptographer Moxie Marlinspike notes that these interactions have become a new primary vulnerability for sensitive information, prompting the creation of tools like Confer that use cryptography to prevent server-side surveillance. For enterprise users, Zero Data Retention policies offer stronger protections by deleting records immediately, though exceptions exist for abuse monitoring and specific high-risk models. Average consumers typically rely on weaker policy-based promises from services like Proton’s Lumo or proxy services like Duck.ai, which do not provide technical guarantees against logging. More robust solutions involve Trusted Execution Environments (TEE) used by Meta and Apple, or local AI tools that keep data entirely on the user's device, albeit with reduced intelligence capabilities. The piece concludes that achieving genuine privacy in AI often requires paying higher subscription fees, as providers cannot subsidize costs through data monetization.
wired