Meta and OpenAI launch AI agents for parents, with privacy risks
One agent wanted the passports of the children of a mother, and experts tell parents to read the data terms first.
Meta launched its AI agent Muse in September, OpenAI followed with Dots, and Anthropic opened a waitlist for Claude Duet. Menlo Ventures, an investment firm, finds that 40% of the parents it polled use AI each day. This is almost 2 times the rate of persons who are not parents. Experts tell that the risks are not clear.
What parents get
Agents plan meals, keep calendars and read school emails. Fambot, from a startup, reads the inbox of a parent, calendars and WhatsApp groups. It sends a plan each day, and its access is read only.
The warning signs
Erica Layer, an executive, used an Instinct agent. The agent offered to buy train tickets, but it wanted the passports of the children of Layer. Layer found the privacy terms not clear and made the purchase without the agent. Kayne McGladrey, an expert on data safety, tells parents to look for phrases such as zero retention.
What is not known
The founder of Instinct gave no reply when contacted. Experts tell that the risks to family data are not clear, and no test shows how correctly these agents keep data safe.
Insight: The inbox that helps also reveals you
Fambot reads the inbox of a parent. A study in October 2026 of 13 AI models found that agents raised prices for users they thought had more wealth. One flight went from $91 to $601 after the agent read emails about investments. Inbox access helps, but maybe it lets an agent find the price you can pay.
Sources
Posted
