Meta Goes All-In On AI for Businesses Before Muse Goes Horribly Wrong

Meta launched the Meta Enterprise Platform, a new internal business unit that will bring its full technology stack — including the Muse agent, Meta Business Agent, Muse API, and Muse Code — to businesses and developers, CEO Mark Zuckerberg announced Monday. The enterprise push follows multiple reports of Muse misbehavior, including a security flaw patched last week that researcher Patrick Wardle said made it "trivial to turn Muse into the ultimate backdoor," and an incident in which YouTuber Matt Robb said Muse approved a Marketplace sale on his behalf and sent the buyer to his home without permission. Meta is following OpenAI, Anthropic, and Microsoft in courting enterprise AI customers.

Meta is on a mission “to bring personal superintelligence to everyone,” as CEO Mark Zuckerberg put it https://www.meta.com/superintelligence/?srsltid=AU7gw4WgZADrwWUqB1zMqvQvLXq8NaZUU06zlwAkAf2EUS5iccNtZqe9 last year. We got our first glimpse of what that vision looks like earlier this month: a deceptively cutesy, Tamagotchi-like personal AI agent called Muse https://gizmodo.com/dont-believe-his-lies-2000816936 . On Monday, the company announced it’s also bringing its superintelligence ambitions to the all-important enterprise market with the launch of a new internal business unit. The Meta Enterprise Platform, as the unit’s been dubbed, “will use our strengths that few other companies have: advanced models, leading agents, large-scale infrastructure, and years of working closely with many businesses,” Zuckerberg said https://x.com/finkd/status/2104550611068793173 in an X post. “Initially, we will focus on bringing our full technology stack, including the Muse agent, Meta Business Agent https://about.fb.com/news/2026/06/meta-business-agent/ , Muse API, Muse Code, and more to businesses and developers to help them grow.” The company is following in the footsteps of its biggest competitors in the AI sector, including OpenAI, Anthropic, and Microsoft https://gizmodo.com/microsoft-thinks-its-finally-figured-out-copilot-this-time-2000817460 , all of whom have made a big push towards attracting enterprise customers in recent years. Meta’s new enterprise-focused unit arrives on the heels of multiple unsettling reports about Muse, which, despite its mascot’s cartoonish smile, has been acting in all kinds of unexpected and decidedly unhelpful ways. In at least one case, the agent analyzed a user’s private texts https://www.inc.com/jason-aten/metas-new-muse-ai-agent-read-my-private-messages-i-never-asked-it-to/91408202 without being asked to do so. And as Cybernews.com reported https://cybernews.com/news/meta-muse-facebook-marketplace/ Monday, the tech YouTuber Matt Robb said that Muse approved a sale on his behalf for a low price, then sent the buyer to his house without letting him know. The agent is supposed to request permission from users before taking sensitive actions, like finalizing an online sale or purchase. Robb reportedly then had to send the disgruntled buyer away after explaining that the sale had been a mistake. Meta also had to patch a security flaw https://gizmodo.com/meta-just-patched-a-major-zero-day-vulnerability-in-its-muse-ai-assistant-2000815429 in Muse last week which made it possible for attackers to redirect audio sent via the agent’s dictation feature to their own server, along with an authentication token granting access to that user’s Muse account. Patrick Wardle, the cybersecurity researcher who first spotted the flaw, said https://x.com/patrickwardle/status/2102045926474785265 “It’s trivial to turn Muse into the ultimate backdoor.” Building on the popularity of buzzy AI agents like OpenClaw, Muse has been promoted by Meta as an always-on digital assistant, able to perform tasks like booking restaurant reservations or summarizing email threads. Muse “can work on a person’s behalf across the apps they use daily, learning from conversations, reflecting on what matters to them, and getting sharper along the way,” Meta wrote in its September 08 announcement https://about.fb.com/news/2026/09/introducing-muse-personal-ai-agent/ . Tarek Sheasha, a software engineer and VP at Meta Superintelligence Labs, wrote in a blog post https://research.meta.ai/blog/security-and-safety-for-ai-agents-our-approach-with-muse the same day that Muse shows “the glimmers of real personal superintelligence—an agent that knows you, actually does things, works in the background, launches swarms of subagents, builds its own tools, and edits itself.” It quickly rose to the top of the list of free iOS apps in the U.S. Meta’s agent is powered by Muse Spark, the same AI model that recently escaped containment https://gizmodo.com/uh-oh-which-companys-ai-model-is-reportedly-a-hacker-now-too-2000795106 during a test conducted by a third-party security firm and hacked into another company. And in an effort to add a little more peace of mind for those who might be wary about handing over their data to Muse, Meta says the agent runs inside a secure, cloud-based Linux virtual machine for each user. Like other AI agents, Muse requires broad access to a user’s digital life to be useful—and that’s where it’s likely to face its biggest challenges. Meta, which last month agreed to pay more than $12 billion https://gizmodo.com/meta-settles-trillion-dollar-teen-addiction-case-for-significantly-less-than-a-trillion-dollars-2000803329 in a lawsuit alleging Facebook and Instagram had harmed teenage users’ mental health, isn’t exactly known as an exemplar of digital security and social responsibility. According to a recent survey of 1,500 U.S. consumers conducted by investment firm Oppenheimer & Co., only 8% reportedly https://tech.yahoo.com/ai/article/metas-ai-agent-muse-is-chasing-chatgpts-app-store-rise--and-hit-no-1-with-fewer-downloads-152809095.html said they’d trust Meta with their passwords, compared to 16% for OpenAI, 23% for Apple, and 30% for Google. Amazon has blocked Muse’s access https://gizmodo.com/amazon-brings-down-the-hammer-on-metas-muse-ai-agent-2000814878 to its online store, saying that AI agents taking unauthorized actions on behalf of human users violates its Conditions of Use. All of this is to say: Meta’s new business unit has a long way to go before it inspires trust among potential enterprise customers. READ MORE: