Meta Builds Photorealistic AI Zuckerberg for Staff Reviews

Compiled by the editorial desk with reference to the Financial Times report and Meta's public statements.

Meta is developing a photorealistic, AI-powered 3D replica of CEO Mark Zuckerberg designed to converse with employees and provide feedback, according to a report from the Financial Times. The initiative, which Zuckerberg is personally overseeing, marks an escalation in the company's use of AI for internal management, though it raises questions about resource allocation and the effectiveness of such tools.

The project is part of Meta's broader strategy to create AI avatars based on public figures that users can interact with in real time. However, the concept has yet to gain traction with the public, as evidenced by the company's earlier foray into character chatbots, which were shut down after less than a year due to poor reception and controversial outputs.

Internal sources told the FT that the AI Zuckerberg could become a significant drain on computing power, which is already in high demand across the company. There are also concerns about the potential misuse of sexualized AI avatars, a risk that Meta has not fully addressed.

The faux Zuckerberg will be trained on a wealth of imagery and voice recordings of the CEO. Zuckerberg is reportedly dedicating five to ten hours a week to what insiders call "vibe coding" to oversee the project personally.

This is not Meta's only AI-driven internal initiative. Staffers are also working on a separate, reportedly unrelated project to develop a "CEO agent" that would help employees retrieve information more quickly, per the FT.

Past AI Avatar Failures

Meta's previous attempts to recreate public figures using AI have not been successful. In October 2023, the company paid celebrities millions of dollars to turn them into chatbots. Following widespread criticism and the chatbots making questionable claims on behalf of their real-life counterparts, Meta shut down the project in less than a year. Yet, some of those chatbots continued to make eyebrow-raising comments well into 2025.

The latest push into photorealistic avatars comes as Meta faces intense pressure to remain competitive in the AI race. The company's Superintelligence Labs recently released its first "Muse Sparks" AI model, which is designed to be computationally efficient and fast. However, in practice, it has underperformed compared to rival models, according to industry benchmarks.

Meanwhile, Meta is pushing employees to adopt AI tools aggressively. During a January earnings call, Zuckerberg told investors that "Meta can get more done" by "investing in AI-native tooling," while "elevating individual contributors and flattening teams."

Last month, reports emerged that Meta was planning sweeping layoffs that could affect at least 20 percent of the company. According to the FT, product managers are running staffers through AI-focused "skills baseline" and "vibe coding" exercises, fueling fears about job security.

For employees who remain, the prospect of receiving feedback from a digital stand-in of Zuckerberg adds a new layer of uncertainty. As the company continues to cut jobs, the introduction of an AI CEO avatar for performance reviews could lead to confusing and unsettling interactions, raising broader questions about the role of AI in workplace management.