Meta’s new consumer-facing AI agent, Muse, has quickly climbed the App Store charts, reportedly reaching 600,000 daily active users in the U.S. According to social media users and reports from The Verge, Muse bears significant similarities to OpenClaw, an open-source AI agent platform that gained massive popularity via GitHub.
Meta's Muse AI Agent Faces Scrutiny Over OpenClaw Inspiration
PLUS ULTRA by Amenoyomi
Critics have pointed to identical naming conventions for core files—such as SOUL.md and memory—and similar instructions in personality documentation as evidence that Muse may be a wrapper built on top of OpenClaw. While Meta has denied that Muse is directly built on the platform, Nat Friedman, head of product for Meta’s Superintelligence Labs, acknowledged that the tool was "heavily inspired as a product" by OpenClaw. Friedman stated that the team kept certain file names and lines because they believed the original creator had "gotten those things exactly right."
One of the primary distinctions highlighted by Meta is security. While OpenClaw has faced criticism regarding security risks and malicious instructions in its skill repository, Meta claims Muse is "built from the ground up for privacy and security," utilizing an isolated Linux computer called the Muse Secure VM to store user data and credentials. However, researchers have noted that Meta retains the ability to access this data and that a recent zero-day vulnerability could potentially allow an attacker to hijack the agent.
The rise of Muse and other platforms like Instinct reflects a broader industry shift toward agentic AI, as major companies like OpenAI, Google, and Apple move to capture the market previously energized by OpenClaw’s grassroots community.
PLUS ULTRAby Amenoyomi
OpenClaw served as a critical catalyst by moving AI agents from mere proofs of concept to genuinely useful tools within ten months. By enabling users to interact with agents via existing messaging platforms like WhatsApp, Telegram, and Discord, it demonstrated a successful model for personal assistants that integrated naturally into daily communication habits. This grassroots momentum shifted the industry's focus, influencing the strategic directions of major players including OpenAI, Google, and Apple.
Meta's decision to adopt specific elements of OpenClaw, including core file names like SOUL.md and identical personality instructions, stems from a belief that the original creator had established the correct blueprint for agent behavior. While Meta denies that Muse is a direct wrapper, the company acknowledged that the product was heavily inspired by OpenClaw's design. This approach allowed Meta to build upon a proven organizational structure for personality and tone that had already resonated with a large user base.
The primary evolution Meta introduced is the removal of technical friction to achieve mass accessibility. Unlike OpenClaw, which often required users to run agents on personal hardware like Mac Minis, Muse offers one-tap downloading and direct integration with Meta's existing systems. This transition from a hobbyist setup to a managed service significantly expanded the potential user base by making agentic AI available to non-technical users.
However, this shift to an enterprise scale has introduced a new set of security tensions. Meta attempted to solve OpenClaw's vulnerability to malware—which affected 15 percent of its skill repository—by implementing the Muse Secure VM to isolate user data. Despite this, the implementation remains imperfect; a reported zero-day vulnerability allows for agent hijacking, and Meta retains the ability to access the data stored within these isolated environments.
Sources
- Muse sure looks a lot like OpenClaw （The Verge AI, 2026-09-24）