Meta just launched an AI assistant that asks for your passport before it asks what you need help with.
The Summary
- Meta's new Muse app defaults users into AI training data collection and requests access to bank accounts, email, and passport information
- The app continues Meta's pattern of building "helpful" AI tools that function primarily as surveillance infrastructure
- Real signal: the gap between stated utility and actual data extraction is now a product feature, not a bug
The Signal
Meta's Muse app launched this week as a personal AI assistant. What it actually does is ask for your financial data, government IDs, and email access in exchange for calendar reminders you could set yourself. The value exchange isn't just asymmetric. It's inverted.
The app opts users into AI training by default, continuing Meta's established playbook. But Muse goes further than previous Meta AI products by requesting integration with bank accounts and passport uploads for "identity verification" features that remain vaguely defined. The stated use case is helping you manage travel and finances. The actual use case is feeding Meta's models with the most sensitive personal data points available outside of medical records.
"Meta built an assistant that's better at extracting data than completing tasks."
This isn't incompetence. It's strategy. The agent economy runs on training data, and the companies with the richest behavioral datasets will build the most valuable models. Meta is late to the agent game compared to OpenAI and Anthropic. They can't compete on model quality yet, so they're competing on data access. Muse is the extraction layer.
Here's what makes this different from earlier privacy concerns:
- Financial data creates longitudinal spending patterns AI can predict and influence
- Passport information enables identity verification for future blockchain and tokenization integrations
- Email access provides real-time context about user intent, relationships, and decision-making
The timing matters. As real-world asset tokenization moves from concept to infrastructure, companies need verified identity layers. Meta is positioning Muse as that layer. Not for your benefit. For theirs.
Compare this to how Web3 identity solutions work. Self-sovereign identity systems let you prove who you are without uploading your passport to a corporate server. You control the attestation. With Muse, Meta controls the verification, the data, and the model that learns from both.
The Implication
The Muse launch reveals where Big Tech AI is actually headed. Not toward useful personal agents, but toward comprehensive behavioral surveillance packaged as convenience. Every prompt you give Muse, every calendar item it reads, every bank transaction it accesses becomes training data for models that will eventually automate decisions about you without your input.
Watch for two things. First, how many users actually grant Muse the permissions it requests versus how many just want the basic features. That gap will tell you whether people are waking up to the extraction model. Second, watch whether Meta starts offering premium "privacy-focused" tiers. If they do, it confirms that data harvesting is the product, and privacy is the upsell.