Meta has released Muse Spark 1.1, an upgraded AI system that the company says can execute real-world tasks rather than simply generate text or images — the latest salvo in a rapidly escalating race to build autonomous digital agents.
Unveiled in mid-July, Muse Spark 1.1 builds on the social media giant's existing Muse image and text tools by adding what Meta calls "action capabilities." The system can browse websites, fill out forms, schedule calendar appointments, and interact with third-party applications through a standardized agent interface — all from natural language prompts.
The release comes as OpenAI and Google have both accelerated their own agentic AI roadmaps. OpenAI's GPT-5.6 family includes "Work" and "Codex" modes designed for multi-step task execution, while Google has been testing agent features within its Workspace suite. Meta's move is notable because it ties agentic AI directly to its massive consumer and advertising ecosystem, including Instagram, Facebook, and WhatsApp.
"Meta is not building an AI assistant — it is building an AI workforce for its platforms," said an analyst at a technology research firm. "If Spark can reliably handle advertiser workflows, content moderation tasks, or e-commerce operations, the operational savings would be measured in billions."
Meta has also integrated Muse Spark into its AI Marketplace tools, allowing businesses to generate ad creative, adjust targeting parameters, and manage campaign budgets through conversational commands. Early testers reported mixed results: the system handled straightforward tasks well but struggled with edge cases requiring human judgment, such as nuanced brand safety decisions.
The company has positioned the update as part of a broader push into what chief executive Mark Zuckerberg has called "the next computing platform." Meta is spending heavily on AI infrastructure, including a planned 1-gigawatt data center in Alberta, Canada, and custom silicon manufacturing set to begin in September.
Privacy advocates have raised concerns about giving an AI system broad access to user data and external services. Meta said Spark 1.1 operates under a permission framework that requires explicit user consent for each action category, though critics argue that most users will accept default settings without reading the fine print.
The agentic AI race is also drawing regulatory scrutiny. The European Union's AI Act, entering full enforcement on August 2, classifies certain autonomous systems as high-risk and imposes strict transparency and human-oversight requirements. Meta has said it is working with EU authorities to ensure compliance, though the final contours of how the law applies to consumer-facing agents remain unclear.
Image source: i.ibb.co