Less than fourteen days after agreeing to an $18 billion multistate settlement regarding social media’s impact on children, Meta has introduced Muse, its most ambitious consumer AI project to date. Announced on Tuesday, the new personal AI agent is designed to handle everyday tasks and projects for users in the United States, requiring a significantly higher degree of consumer trust than the company’s social platforms ever did.
To function, Muse connects to a wide array of services integral to daily workflows, including email, calendars, payment systems, fitness apps, smart home devices, dining, shopping, music, and event platforms. The agent marks a strategic shift from the current ChatGPT era of conversational bots toward actionable assistance. According to Meta, Muse can send emails, book travel, negotiate lower bills, fill out forms, create plans, convert recipe videos into grocery lists, send invitations, and execute purchases via Stripe’s Link service. Additional integrations with Shopify’s Shop Pay and 1Password are expected soon.
Muse operates on Meta’s Muse Spark model and features built-in connectors for various services, with more planned. Users can selectively opt into specific app connections one at a time. If a desired service lacks a public API, the agent can establish a connection using user-provided credentials or access it through a browser interface.
The service will launch free of charge via web at muse.ai, along with iOS and Android apps, and through WhatsApp chats. Support for Meta’s AI glasses is also anticipated. While the base service is free, paid tiers will follow as usage increases. Two subscription plans are available at launch: Power at $20 per month and Maximum at $100 per month. Meta states that most users will likely remain on the free tier, supported by a usage meter that alerts users before they hit payment thresholds.
Like other emerging AI agents, Muse continues to operate in the background and improves over time by learning user preferences to offer unprompted suggestions. The agentic AI space is becoming increasingly competitive, with companies like Google and Anthropic releasing tools such as Gemini Spark and Claude Cowork, while Apple integrates AI into iMessage and SMS.
However, the rise of autonomous agents has intensified scrutiny over privacy. Early testers of competing assistant Instinct expressed shock upon discovering the app demanded a broad license to use and store user data for AI training. In response, Meta claims Muse runs within a dedicated, secure virtual machine equipped with its own browser. A separate Sentinel agent operates on the same infrastructure but is systemically isolated from Muse, ensuring the primary agent cannot access passwords or payment methods. Meta also asserts that conversation data is not shared with its advertising systems, though these claims require independent verification by security experts.
Despite these technical safeguards, Meta faces significant hurdles in building consumer confidence. The company has a lengthy history of regulatory conflicts, including a 2011 FTC settlement for deceptive privacy practices, a record-breaking $5 billion penalty in 2019 for multiple privacy violations, and a 2023 FTC charge for violating a previous privacy order. Technical missteps, such as the 2019 revelation that user passwords were stored in readable formats and the pervasive Cambridge Analytica scandal, continue to affect public perception.
Furthermore, Meta has faced repeated congressional testimony regarding the protection of minors, leading to recent legal victories for plaintiffs. In addition to the $18 billion settlement with 29 states, a New Mexico court ordered Meta to pay $942 million and overhaul child safety protections, while thousands of other personal injury and school district cases remain pending across the industry.
To mitigate consumer fears, Meta has provided extensive technical documentation and introduced customizable features allowing users to name their agent, select avatars, and adjust communication settings. Whether these personalization efforts and utility gains will be sufficient to restore trust remains to be seen.
Leave a Reply