Meta is building an AI double of Mark Zuckerberg so employees can talk to the boss

Meta is developing a photorealistic AI version of Mark Zuckerberg that employees can use to discuss company strategy and ask questions, instead of talking to the CEO in person. The project is run by Meta Superintelligence Labs, and Zuckerberg himself takes part in training and testing. Separately, the company is building a CEO agent, an internal assistant for the executive himself. No launch date for the digital double has been disclosed. *Meta has been designated an extremist organization and is banned in Russia.

AI-processed from TNW; edited by Hamidun News
Meta is building an AI double of Mark Zuckerberg so employees can talk to the boss
Source: TNW. Collage: Hamidun News.
Listen to article
Article Fact Sheet & Key Figures
ParameterDetails
Primary sourceTNW ↗
Editor-in-Chief
Key metric #1A photorealistic avatar is being created for conversations with employees, not for an external audience
Key metric #2The model is trained on Zuckerberg's manner of speaking, tone and strategic views
Key metric #3The project is at an early stage, and there is no launch date yet

According to the Financial Times, Meta is developing a photorealistic AI double of Mark Zuckerberg for internal conversations with employees. The idea is that some questions about the company's strategy and structure could go to a digital version of the CEO, rather than waiting for a face-to-face meeting with him.

Why build a double

The project is a separate initiative inside Meta and is not the same as another internal tool, the so-called CEO agent, which is meant to help Zuckerberg himself get information faster and handle work tasks. The new product is conceived differently: it is not an assistant for the executive but a digital conversation partner that can talk to people on his behalf.

The avatar is trained on Zuckerberg's manner of speaking, his tone, his public appearances and his views on company strategy. According to sources, Zuckerberg himself takes part in training and testing the system.

For Meta, this is an attempt to solve a simple but painful problem of scale: the company has almost 79,000 employees, and few of them have direct access to the founder. If the digital version can answer routine questions about priorities, products and the direction of the business, Meta gets a new internal communication channel without the constant involvement of the real CEO.

"Employees should feel closer to the founder through this kind of conversation."

How Meta got here

The work is being done by the Superintelligence Labs division, where Meta is building more realistic AI characters capable of holding a conversation with almost no delay. This is where the main technical difficulty lies: it is not enough to generate a similar face and voice; the system also needs lifelike facial expressions, natural speech and fast real-time reactions. That level of believability requires serious computing infrastructure, so the project looks less like a demo experiment and more like an expensive internal bet on a new communication interface.

Zuckerberg has reportedly become more deeply involved in Meta's AI development in recent months: he spends five to ten hours a week on code, takes part in engineering reviews and promotes the idea of personal superintelligence. Against that background, an AI clone looks like a logical extension of a course toward a company where algorithms not only help write code or search for data but also replace part of management communication inside a huge corporation every day.

  • A photorealistic avatar is being created for conversations with employees, not for an external audience
  • The model is trained on Zuckerberg's manner of speaking, tone and strategic views
  • The project is at an early stage, and there is no launch date yet
  • Separately, Meta is building a CEO agent that helps Zuckerberg himself rather than replacing him
  • The goal of the project is to shorten the distance between the founder and a large organization

Meta's interest in the format did not come out of nowhere. In 2023 the company launched chatbots with celebrity likenesses, including Snoop Dogg, Tom Brady, Kendall Jenner and Naomi Osaka, but in summer 2024 that lineup was quietly shut down: it had no noticeable impact. Meta then opened AI Studio for creating custom characters, but quickly ran into the problem of explicit and sexualized bots, and since January it has restricted teenagers' access to AI characters. The success of Character.AI with younger audiences also reportedly influenced interest in the current project.

What this means

If Meta gets the project working, large companies will have a new template for communication between leadership and teams: not rare all-hands calls but a permanent AI channel with a "version" of a top executive. This could speed up answers to internal questions and make strategy clearer, but it also raises an uncomfortable question: where does a convenient interface end and the replacement of live leadership with its simulation begin. For corporate AI, this is no longer a toy but a new organizational role.

*Meta has been designated an extremist organization and is banned in Russia.

Story

Key Questions Answered

Why build a double?

The project is a separate initiative inside Meta and is not the same as another internal tool, the so-called CEO agent, which is meant to help Zuckerberg himself get information faster and handle work tasks. The new product is conceived differently: it is not an assistant for the executive but a digital conversation partner that can talk to people on his behalf. The avatar is trained on Zuckerberg's manner of speaking, his tone, his public appearances and his views on company strategy. According to sources, Zuckerberg himself takes part in training and testing the system. F

How Meta got here?

The work is being done by the Superintelligence Labs division, where Meta is building more realistic AI characters capable of holding a conversation with almost no delay. This is where the main technical difficulty lies: it is not enough to generate a similar face and voice; the system also needs lifelike facial expressions, natural speech and fast real-time reactions. That level of believability requires serious computing infrastructure, so the project looks less like a demo experiment and more like an expensive internal bet on a new communication interface. Zuckerberg has

What this means?

If Meta gets the project working, large companies will have a new template for communication between leadership and teams: not rare all-hands calls but a permanent AI channel with a "version" of a top executive. This could speed up answers to internal questions and make strategy clearer, but it also raises an uncomfortable question: where does a convenient interface end and the replacement of live leadership with its simulation begin. For corporate AI, this is no longer a toy but a new organizational role. Meta has been designated an extremist organization and is banned in

ZK
Zhemal Khamidun — CPO at Alpina Digital, product manager of AlpinaGPT. Building AI products since 2019. Hamidun News is a pet project that grew into a production aggregator with 6,000+ readers.
✦ Unified AI Workspace · Nomad

Test Frontier LLMs & Autonomous Agents in One Workspace

Run Claude Opus 5.5, GPT-6, Gemini 3.1, and DeepSeek side-by-side in Nomad AI (nomadnet.ai) — unified model routing, deep research, and multimodal workflows under one subscription.

Try Nomad AI →Launch Nomad Agent
What do you think?
Loading comments…