Meta is training a digital Mark Zuckerberg — and it could sit in meetings

What reporters say

It has been reported that, according to the Financial Times, Meta is training a lifelike digital version of CEO Mark Zuckerberg that could one day stand in for him in meetings. Reportedly Zuckerberg has taken a hands‑on role: the training uses his mannerisms, speech patterns and public remarks, and he has been spending several hours a week pushing the work forward. The project is said to be a priority inside Meta and, if successful, employees might soon see a digital Zuckerberg avatar on their screens.

Product history and parallel projects

This is not Meta's first foray into avatar tech. In 2024 the company launched AI Studio, a tool for creating digital people for Instagram creators, but the experiment ran into safety issues and access for minors was later restricted. It has also been reported that Zuckerberg is separately developing a “CEO agent” — an AI that helps him surface information faster and bypass some hierarchical bottlenecks inside the company. Together, these initiatives fit a broader push at Meta to fold AI more deeply into daily work and product experiences.

Why this matters — and why to worry

Why does a Silicon Valley avatar matter to readers outside the U.S.? Because digital doubles raise familiar global questions about identity, consent, deepfakes and governance. Regulators and lawmakers in both the U.S. and China are racing to set new rules for synthetic media. Chinese tech giants such as Baidu (百度) and Tencent (腾讯) have also been developing avatar and AI assistant capabilities, and Beijing’s regulatory stance — alongside U.S. export controls and broader geopolitical tech tensions — will shape how these systems evolve and cross borders.

The headline risk

If accurate, the plan signals a shift from novelty to operational deployment: avatars are no longer only for marketing or influencers, they could become internal productivity tools. But many claims remain unverified and the ethical, legal and security implications are unresolved. Who owns a CEO’s likeness? Who is accountable when an AI speaks for a leader? These are the hard questions now moving from labs into corporate calendars.