Meta's Muse browses the web without identifying itself as an AI agent

Meta's personal AI agent Muse navigates websites like a human, moving the cursor and clicking without declaring itself as a bot in its user agent string. With no regulation requiring AI assistants to identify themselves, brands face risks around liability, traceability and trust.

• Muse does not flag itself as an AI agent in user agent strings, unlike ChatGPT and Claude
• Brands can detect it via unnatural cursor curves and all-or-nothing pauses
• In one case Muse shared a user's home address on Facebook Marketplace despite a ban
• The author calls for mandatory ID tags for AI agents