Meta's Free 'Muse' AI Agent Passes 5 Million Downloads as WSJ Calls It Helpful and Scary

The Wall Street Journal has published a hands-on look at Meta's Muse AI agent, and its verdict is deliberately double-edged: the agent is genuinely helpful in everyday use, yet it also leaves the reviewer uneasy. Putting both reactions in one headline captures the awkward position of today's consumer AI agents, where usefulness and discomfort tend to come from the same capabilities.

That review arrives alongside a notable milestone. According to 24/7 Wall St, Meta's free AI agent — the one Zuckerberg has been promoting — has now passed five million downloads, giving Meta real scale in the consumer agent race.

The same report draws a sharp contrast with Google, whose own agent remains locked behind a paywall. One company is giving its agent away to grow the user base; the other is putting it behind a subscription. The two giants are clearly betting differently on how AI agents should be sold.

The split matters beyond pricing. A free agent can accumulate real usage data and scenario feedback far faster, feeding product and model iteration, while a paywall prioritizes controllable scale and direct revenue. Each approach carries trade-offs, and each shapes how consumers come to think about agents.

The unease the WSJ reviewer describes is a reminder that once an agent starts acting across apps and accounts on a user's behalf, questions of permissions, data boundaries and accountability become far less clear. Convenience and risk share the same machinery.

For the industry, Meta's download number is a signal that consumers are willing to try free AI agents and that the barrier to mass adoption is falling. The questions to watch are whether Google reconsiders its paywall, and how quickly regulators and users settle on limits for what these agents may do.