How to safeguard your privacy while using AI

AI chatbots like ChatGPT, Claude, and Gemini have become integral to personal life, often collecting sensitive data without clear restrictions. Matt Green, a computer science professor, warns that users are essentially revealing intimate details to these models. Moxie Marlinspike, creator of Signal, highlights that privacy risks have shifted from messaging to AI interactions. He launched Confer, an AI chatbot that uses cryptography to prevent server access to user conversations. Other services, like Lumo and Duck.ai, promise not to log conversations but relay data to third-party models. Privacy-focused tools like Confer and Meta’s TEE system aim to offer stronger protections, though many consumer services lack robust data retention policies. Apple’s Private Cloud Compute and Meta’s Muse Confidential VM are among the latest efforts to enhance privacy in AI interactions. Despite these measures, most AI services still pose privacy risks, as data can be collected and shared. Experts emphasize that technological guardrails, like trusted execution environments, are crucial for ensuring user data remains private. While some companies offer zero data retention policies for enterprise users, average consumers face limited protections. The growing competition among AI providers aims to address these concerns, but users must remain vigilant in choosing services that prioritize privacy.