Meta launches Muse, a personal AI Agent that can do your everyday tasks; company makes a 'promise' after Muse AI Image 'fiasco' that allowed anyone to create AI images with users' Instagram photos

Meta has launched a new autonomous artificial intelligence (AI) assistant capable of executing tasks such as sending emails, listing a car for sale, and booking travel. The system, dubbed "Muse" and known internally as "Hatch," reportedly serves as the cornerstone of CEO Mark Zuckerberg’s ambitions to deliver what he calls "personal superintelligence" across Meta's global platform. In a post on X, Meta CEO Mark Zuckerberg wrote, "Introducing Muse, the personal agent that understands your goals and works 24/7 to get things done for you."Initially limited to U.S. users, Muse will operate through a standalone app and within Meta’s WhatsApp messaging platform. The company that for now, Muse launches on iOS, Android and the web, with free and paid tiers, and support for AI glasses coming soon.Based on the open-source OpenClaw framework, Muse functions by linking directly to a user’s calendar, email, financial services, smart home hardware, and health platforms. Meta noted that users must manually grant permissions for each service and retain the ability to revoke access at any point.Unlike traditional chatbots, each Muse instance runs within an isolated, cloud-based virtual machine designed to emulate an individual personal computer. This architecture allows the software to execute long-running tasks autonomously in the background without requiring continuous user interaction.Incidentally, the Muse AI agent comes despite reported internal concerns regarding how the software handles sensitive personal data. The system’s deep integration with personal data has intensified scrutiny over potential safety, privacy, and operational hazards—particularly the risk of the agent executing erroneous transactions or mismanaging private information.Meta's security promiseMeta has made a strong security promise including claims of implementing additional security safeguards. According to Vishal Shah, the company’s vice president of AI products, the engineering revisions successfully brought the tool in line with Meta's baseline thresholds for privacy, security, and model performance. "It is impossible to say that there is never going to be a mistake, but every single part of the architecture has been designed to make this as safe, as secure, as private as we can possibly make it," Shah said. Meta AI chief Alexandr Wang said the app runs within “its own isolated environment” inside the company’s computing infrastructure, and “never sees your actual passwords or payment details.”In July this year, Meta launched a new AI image generator, Meta AI Image, that allowed anyone to create AI images with users' Instagram photos. Also, if users had a public Instagram account, Meta automatically opted them in.Muse comes hours before Apple's Big eventMeta Muse AI assistant arrives just hours before Apple upgrades its voice assistant Siri AI for the chatbot era. While Siri AI is capable of taking some actions, it is not a full blown agentic system yet like OpenAI, Anthropic, SpaceXAI and Meta AI offer. Apple is likely to launch it at its September 9 event, along with new iPhones.Get the latest technology news and updates. Download the TOI App.