OpenAI and Meta promise more privacy in new AI agents

In short: AI agent companies are competing by saying their tools protect your personal data better than rivals.

What's going on

OpenAI used its DevDay event to introduce a new AI agent called Dots. An AI agent is a helper that can take actions for you, like using a browser or handling tasks, not just answering questions. OpenAI CEO Sam Altman said the company wants to “set a new standard for privacy” as these agents become more powerful.

OpenAI also criticized Meta’s competing agent, Muse, suggesting it does not do enough to keep user data safe. But Meta launched Muse only months earlier with a similar message. Meta CEO Mark Zuckerberg said Muse was “built from the ground up for privacy and security,” and positioned it as safer than an earlier agent called OpenClaw.

Meta says Muse keeps user data inside a “secure VM,” which is short for virtual machine (a locked-off computer inside a computer). Zuckerberg described it like an isolated Linux computer that includes a browser, processing power, memory, and storage. Meta also said that a large share of Muse’s development work focused on designing it to operate more safely.

What to watch

These promises matter because AI agents may need access to sensitive information to be useful, like emails, calendars, and websites you log into. It is like giving a personal assistant your keys and expecting them to never copy them. Watch for independent testing, clear privacy policies, and simple controls that let people limit what an agent can see, store, or share.

Source: The Verge AI