If you're sharing your deepest secrets with an AI chatbot, you might want to rethink your approach. Matt Green, a privacy-focused computer science professor, warns that these tools are like a digital confession booth, collecting your most intimate thoughts and data. However, some new AI services are trying to change that by offering zero data retention policies and strong encryption to protect users' privacy.

Moxie Marlinspike, creator of the widely-used encrypted messaging app Signal, has launched Confer, an AI chatbot that uses cryptography to prevent the server from surveilling or logging conversations. This is just one of many new tools designed to protect your privacy in the age of AI. However, even with these safeguards, the big players like OpenAI, Anthropic and Google still retain some level of data and might have exceptions to their privacy policies.

While these services make privacy claims, they often come with significant exceptions. For instance, Anthropic doesn't offer zero data retention for its most advanced models, and OpenAI retains some data for safety reasons. These policies are only available for enterprise and developer accounts, leaving the average user with weaker promises of privacy.

As AI becomes a more integral part of our lives, it's crucial to understand the privacy implications. Whether you're using a service that promises zero data retention or one that only offers weaker assurances, it's important to be aware of the potential risks and limitations. The technology is advancing, but the battle for privacy is far from over.