Brief IA
›

Meta Promotes Muse: Balancing Success and Trust Challenges

AI Tools·Tom Levy·

Meta Promotes Muse: Balancing Success and Trust Challenges

Meta Promotes Muse: Balancing Success and Trust Challenges
Key Takeaways
1Meta has delayed the launch of Muse to enhance security, enlisting Moxie Marlinspike
2The assistant has reached the top of the App Store and Meta's stock has risen by over 20%
3Muse aims for monetization through commissions on purchases, but privacy and competition remain challenges
Why it matters : Muse marks Meta's return to the AI race, but its sustainability will depend on user trust and its ability to stand out against the competition.

Meta has delayed the launch of Muse for security reasons and has enlisted Moxie Marlinspike to co-design its protective architecture. The assistant, which has reached the top of the App Store, hopes for a commission-based transaction model and is receiving a warm welcome on Wall Street, but privacy and upcoming competition remain challenges as Meta prepares a new model named Watermelon.

Security and Privacy Dictate the Launch Pace

Meta has postponed the launch of Muse due to security and product accountability concerns, according to Mark Zuckerberg. The CEO claims to have enlisted Moxie Marlinspike, along with Nat Friedman, to co-design the security architecture of the assistant. The application requires extensive access to users' accounts, raising privacy concerns. By design, an agent like Muse also exposes users to security and privacy risks, including potential unauthorized purchases. These issues are identified as major barriers to adoption by analysts at Bank of America. Users like Hadi Javeed have also raised these privacy questions.

Commission-Based Monetization and Wall Street Reception

Mark Zuckerberg wants Muse to enable users to buy and save money, with Meta planning to take a small commission on each transaction. On Wall Street, Muse is seen as an opportunity for Meta to move beyond its advertising model, a scenario that analyst Lloyd Walmsley believes is unfolding. Since its launch, Meta's stock has risen by over 20%. A Meta spokesperson highlights the service provided to millions of people thanks to the efforts of many employees and asserts that the company continues to make progress.

Public Adoption and Simplified Onboarding Journey

Users praise the ease of getting started with Muse. Hadi Javeed reports that his father saw an advertisement on Facebook, downloaded the app, and quickly purchased a pair of shoes, believing that the setup had been streamlined. After a launch overshadowed by the Instinct agent, Muse has gained traction by leveraging the built-in audience of Meta's applications, including among people who had never used an AI agent before. The assistant has reached the top of the App Store.

Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisissez votre rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

From "Hatch" to Early Concrete Results

Nat Friedman has been working on Muse since September 2025 within the superintelligence unit, with the project having the codename "Hatch." Two weeks before this month's launch, the agent was still struggling with English before performing better on consistency tests. Earlier this year, internal AI experts had deemed it underperforming, likening it to a simple chatbot with a weaker model. Tens of thousands of employees then tested the agent for months, providing use cases that helped improve tasks like scheduling meetings and canceling flight tickets through a "hill-climbing" process. The team also trained Muse to limit the sharing of sensitive information and to use new tools for seamless interaction with the Internet. At the Connect conference, Alexandr Wang highlighted the researchers' efforts to teach agents to choose the right tool during an event largely dedicated to the application.

Targeted Recruitment and Nat Friedman’s Role

Meta has launched a hiring spree for agent-based AI and has made acqui-hires, including the co-founders of Dreamer, David Singleton and Hugo Barra, along with their team. David Singleton, who joined in March, is working on Muse and has helped design the core technology for securely storing sensitive information. Zuckerberg also brought in Nat Friedman last year as part of a deal that included a stake in his NFDG fund. Friedman and Alexandr Wang had co-authored a memo describing a "superintelligent personal assistant." Internally, Friedman’s decisive style may put him at odds with other teams.

Active Competition and Meta's Roadmap

The market remains dynamic, with players like Instinct thriving and an OpenAI product expected soon. Earlier this year, OpenAI hired Pete Steinberger, creator of OpenClaw. At Meta, the fear of being outpaced has been strong, with the team surprised that months went by without a major new entrant. Looking ahead, Meta is working on an AI model named Watermelon and claims it has already caught up with OpenAI. Alexandr Wang indicates that other ideas "as big" as Muse are in the works and posted on X that "the kitchen is just starting to heat up." However, Meta's ability to maintain the momentum of Muse remains uncertain.

Brief IA — L'actualité IA en français

L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.