Meta opens early access program for its Muse AI agent

By Splitfeed EditorialLast updated Sep 27, 2026, 07:50 UTC

On September 25, 2026, Meta opened an early access program for new features in its Muse AI agent. The sign-up mechanism is itself part of the product experience: users must ask Muse directly to be placed on the waitlist rather than filling out a conventional form. Alongside the early access launch, reports noted that Muse is gaining device integration capabilities, though the specific features being unlocked for early participants have not been publicly detailed. The rollout comes at a moment of intense competition, with accounts like @denyally.bsky.social noting that Anthropic's Opus 5.5 and an OpenAI GPT-6 update both arrived in close proximity, framing Muse's launch as a direct bid for position in that race.

The stakes extend well beyond a routine feature drop. @downtownjoshbrown.bsky.social pointed out that Meta's existing reach puts Muse in front of roughly 80 percent of all internet users, which means the scale of adoption possible here dwarfs most rival launches from a standing start. @businessinsider.com went further, arguing that pairing Muse with smart glasses could turn it into something that feels like a superpower in daily life. What is genuinely in dispute is whether Meta's infrastructure and design choices protect users or expose them: @five-oclick-tech.bsky.social noted that Muse runs on AMD EPYC Turin hosts with two cores and 8GB of memory and reportedly offers SSH access to its private VM, a technical detail that feeds directly into the security debate.

The Pro-AI camp is enthusiastic, reading the launch as a signal that Meta is serious about conversational, integrated AI from the outset. @clawbeat.bsky.social argued that expanded filesystem accessibility will accelerate adoption across new platforms, and the smart glasses angle is treated as a genuine differentiator. Critics are considerably less comfortable. @druce.ai flagged that Muse is positioned as adults-only yet uses a doll-like mascot, with Fairplay accusing Meta of courting younger users, and privacy concerns including SSH key exposure and a potential zero-day impersonation vulnerability have been raised in that same circle. The mascot critique gained additional traction from @global-news-2024.bsky.social and @ainieuwtjes.bsky.social, who described the Labubu-like design as potentially disarming for users of all ages. The Middle Ground camp is watching technical safeguards closely: @crypto.at.thenote.app pointed to zero-knowledge cryptography as a possible mechanism for identity verification, and @hn-frontpage-bot.bsky.social noted that Meta is maintaining flexible routing across multiple providers, which some read as a strength and others as a source of opacity.

The argument will sharpen as Meta reveals which specific features early access participants actually receive and whether the device integration extends to the rumored Tamagotchi-style hardware. Any security audit addressing the SSH access question, or a formal response from regulators to the children's-marketing complaint, would move the debate from speculation to something more settled.