Meta Admits Muse AI Agent Was Heavily Inspired by OpenClaw After Users Spot Identical Files

Meta Confirms OpenClaw Inspiration Behind Muse AI Agent

Meta Superintelligence Labs head Nat Friedman publicly acknowledged that Meta Muse was "definitely heavily inspired as a product by OpenClaw," addressing weeks of speculation from early adopters who noticed striking similarities between the two platforms1

1

. The former GitHub CEO, who joined Meta last year alongside Chief AI Officer Alexandr Wang, revealed that after using the open-source AI agent OpenClaw in January, he purchased hundreds of Mac minis for his team at Meta Superintelligence Labs3

3

. "Lots of us fell in love with using OpenClaw," Friedman wrote on X, explaining that the goal with Muse was to build something like OpenClaw that could be made safe, secure, easy to use, and scalable to billions of people4

4

.

Users Discover Identical System Configurations

The controversy erupted when AI app co-founder Ansh Nanda posted a viral thread claiming "Muse is LITERALLY OpenClaw for normies"1

1

. Nanda included a conversation with the Muse AI agent itself, where it acknowledged that similarities between its system files and OpenClaw's were "not just a coincidence" but rather "a match"5

5

. Users quickly discovered that both platforms use identical names for core files including SOUL.md, memory, and tools. The SOUL.md file, a plain text configuration file written in Markdown that defines an AI agent's personality, communication style, tone, values, behavioral boundaries, and expertise, contained nearly identical content between the two platforms2

2

. Both files included similar phrases like "Be genuinely helpful, not performatively helpful"2

2

.

Friedman Defends Design Choices

When questioned about why Muse copied the exact file names and had nearly identical content in the SOUL.md file, Friedman didn't dispute the claims. Instead, he simply stated that "we thought that Peter got those things exactly right"1

1

. He called Peter Steinberger, the Austrian developer who created OpenClaw, "a genius" and described his harness as "pioneering from the jump"3

3

. The harness refers to the scaffolding around a model that holds instruction files, tool definitions, and the loop deciding what the agent does next3

3

. While Friedman emphasized that Muse was built from scratch, the acknowledgment highlights Meta's well-known playbook of taking promising products and copying their best features, most notably with Snapchat's Stories format1

1

.

OpenClaw's Revolutionary Impact on Personal AI Agents

OpenClaw transformed consumer-facing AI agents from proof of concept to genuinely useful tools in just 10 months2

2

. Created as a one-man weekend project by Steinberger and running off users' personal computers, it allowed people to converse with their AI agents through messaging platforms like WhatsApp, Telegram, Slack, Teams, or Discord2

2

. Within about a week, the open-source AI agent garnered 2 million visitors and 100,000 stars on GitHub, spurring people to buy Mac minis to run their agents 24/72

2

. Major AI companies took notice: OpenAI hired Steinberger in February to work on agents, Google announced its consumer AI agent bets in May, Apple went all-in on agents in June, and Instinct raised hundreds of millions of dollars in August at a $2.5 billion valuation2

2

.

Muse's Rapid Market Success Raises Security Questions

Meta launched Muse on September 8, and it quickly reached No. 1 on the U.S. App Store3

3

. Data indicates the app is outpacing ChatGPT's launch when platform and market availability are compared directly1

1

. More than 500,000 people tried it in roughly its first week, including over 250,000 daily users who submitted upwards of 2 million prompts3

3

. The Muse AI agent handles tasks like booking appointments, filling in forms, customer service, purchases, image generation, and document creation, requiring access to email, calendar, messaging, social accounts, and macOS permissions3

3

. Meta offers the app free within limits, with paid tiers at $20 and $100 per month3

3

. However, security concerns persist. While Zuckerberg claims Muse is built from the ground up for privacy and security with user data stored on the Muse Secure VM, Meta can still access this data2

2

. The company plans to introduce cryptographic verification to prevent Meta from accessing data later this year, but a zero-day vulnerability flagged by researchers allows attackers to hijack the agent and take complete control2

2

. Watch for how Meta addresses these security gaps and whether other platforms like Instinct and Google Home continue integrating OpenClaw-style frameworks into their consumer-facing AI products.