Remember Mark Zuckerberg’s awkward Horizon Worlds avatar—the blocky, legless selfie that became the internet’s favorite punchline? That cartoonish misstep may soon be a relic of Meta’s past. According to the Financial Times, Meta is building a photorealistic, AI-driven 3D clone of Zuckerberg intended to hold real conversations with employees on his behalf. How it works: the system is being trained on Zuckerberg’s mannerisms, vocal patterns, public remarks and recent strategic thinking so it can “talk like him, think like him” and, in Meta’s words, “make employees feel more connected to the founder.” Four people familiar with the project told FT that Zuckerberg is personally involved in training and testing the model. Who’s building it: the work is led by Meta’s newly formed Superintelligence Labs. Scaling a lifelike, low-latency conversational agent demands enormous compute, which has proved challenging. To bolster audio capabilities, Meta bought two voice startups last year—PlayAI and WaveForms—and has dramatically increased its capital plans: projected capex for 2026 is $115–135 billion, nearly double last year’s figure. Product rollouts and internal push: last week Meta unveiled Muse Spark, the first model from Superintelligence Labs, pitched as a compact, purpose-built model with strengths in health reasoning and visual understanding. The announcement sent Meta shares up about 7%. Inside the company, employees are being encouraged to build agents using an open-source toolkit called OpenClaw; product managers are working through a “skills baseline exercise” that includes system-design tests and even what Meta calls “vibe coding.” A sharp turnaround from the metaverse era: the photorealistic clone marks a pivot away from the clunky virtual-world experiments that defined Meta’s earlier strategy. Horizon Worlds’ cartoon avatar became emblematic of that flop—employees reportedly avoided the platform during a 2022 “quality lockdown,” and Reality Labs’ losses were staggering (Meta’s Reality Labs spent roughly $10.2 billion in 2021 alone), prompting a quieter strategic shift. Why it matters to crypto and Web3 watchers: the move highlights growing industry momentum toward AI-driven identity and agentization—the same forces that can enable novel on-chain interactions, DAO governance automation, or tokenized representation. It also raises familiar crypto-era questions about authenticity, impersonation, and surveillance: does a convincing founder-bot foster connection in a distributed org, or simply extend managerial oversight through a more human-sounding interface? Bottom line: Meta is betting the next phase of internal communication will be hyperreal AI agents rather than plastic metaverse avatars. If successful, the project could reshape how executives scale presence across large organizations—but it also magnifies ethical and security concerns that matter to both centralized tech players and the decentralized communities watching them. Read more AI-generated news on: undefined/news