Meta and OpenAI Promise Private AI Agents — Can You Trust Them?

Meta and OpenAI are competing to win your trust with new AI agents — AI that can do tasks for you, like reading emails or making purchases. Both companies promise their agents are private and secure. Meta's Muse, launched a few months ago, was pitched as a safer alternative to an earlier agent called OpenClaw. CEO Mark Zuckerberg said Muse was "built from the ground up for privacy and security." But soon after launch, a security researcher found a flaw that could let someone take control of Muse. Also, Muse collects your data by default and even reads your private messages without asking, as one reporter found. It also creates detailed profiles of your friends and family.

OpenAI then announced its own agent, Dots, at its DevDay event. CEO Sam Altman said OpenAI wants to "set a new standard for privacy in frontier AI." OpenAI took shots at Meta, saying it has more experience with billions of users and would avoid such mistakes. Dots lets you set rules, like never spending over a certain amount. So far, Dots hasn't had major privacy scandals, but it's only available on ChatGPT plans that cost $100 or more per month, so fewer people use it.

The catch? AI agents need a lot of your personal data to be useful. They read your messages, know your contacts, and can act on your behalf. That means trusting a company with even more of your life. Both Meta and OpenAI say they'll protect your data, but as we've seen, promises can fall short. For now, if you use these agents, be aware of what you're sharing and check your privacy settings. The convenience might come at a cost.