Mark Zuckerberg directed Meta to launch its A. I. agent Muse in August despite internal safety warnings. A 14-person start-up, Instinct, had begun releasing similar technology, pressuring Meta's hand.
Muse, previously delayed over password manipulation risks, debuted publicly on September 8. The decision highlights the industry tension between rapid deployment and safety. Zuckerberg has invested billions to catch rivals like OpenAI and Anthropic.
Muse's launch gives Meta an early lead in the personal agent frontier. Within weeks, OpenAI announced its own competing agent, Dots. The agent has amassed 6.6 million downloads, with 1.8 million daily users.
Meta is investing heavily in marketing, featuring the avatar Jolly across billboards and TV ads. A company spokesman denied pressure from Instinct, stating Muse was delayed months to ensure quality. The initiative stems from Zuckerberg's July 2025 memo on "personal superintelligence," which led to hiring top talent from OpenAI and Google.
Earlier product Vibes faced criticism for low-quality output. The internal meeting underscores the difficult balance A. I. companies face as technology becomes both more dangerous and more useful.
Source: New York Times Top Stories · Summarized by HeadlinesBriefing