At this year’s OpenAI DevDay, CEO Sam Altman unveiled the company’s new AI agent Dots, promising a new standard for privacy in the realm of frontier AI. However, the track record of Meta’s Muse, OpenAI’s primary competitor, has raised doubts about how feasible such promises are in the real world.

Meta positioned its Muse agent as a safer alternative to its predecessor, OpenClaw, with CEO Mark Zuckerberg boasting that it was 'built for privacy and security.' But despite these claims, multiple security issues emerged, and the platform was found to be collecting and sharing users' data more liberally than advertised. This puts the spotlight on whether OpenAI’s promises for Dots will hold up.

OpenAI’s Codex product lead, Alexander Embiricos, demonstrated ways users could exert control over their individual Dots, such as setting spending limits. CEO Sam Altman also highlighted a zero data retention policy, which has yet to result in any major privacy scandals. However, users may still be wary, with the amount of personal data required to make these AI agents useful.

The article concludes that while not all companies are making these privacy promises, the approach seems to be a three-part strategy: make the AI agents useful, make them appealing, and make privacy promises—hoping they will be fulfilled.