Meta is training an AI version of Mark Zuckerberg’s clone
Meta is training an AI clone of Mark Zuckerberg using his voice, image, and public statements. The goal is for the company’s nearly 79,000 employees to be able to consult the “virtual Zuckerberg” when they don’t have access to the real one. According to the Financial Times, the CEO himself is actively participating in training his artificial version.
Meta has been experimenting with digital avatars for years. In 2022, Mark Zuckerberg introduced his avatar in the Metaverse — which was heavily criticized for its graphic quality — and since then the company has been developing AI-generated 3D characters capable of holding real conversations. Now the project has taken a leap forward, with the focus on building a replica based on the CEO himself.
In 2022, Mark Zuckerberg introduced his avatar in the Metaverse — which was heavily criticized for its graphic quality.
Meta believes the experiment could be replicated with influencers and content creators. If this scales, it could change how people and brands communicate with their followers online. Not only that, it could also make it possible to virtually clone anyone to obtain their “opinion” based on prior training.
OpenAI launches a cybersecurity-focused model that it will share only with selected organizations. Just days ago, Anthropic announced it had created Mythos, a cybersecurity-focused AI to which it granted access only to 40 organizations it considered “good actors,” such as Apple, Amazon, Microsoft, Google, and the Linux Foundation. A week later, OpenAI announced it would do something very similar with GPT-5.4-Cyber, initially distributing it to hundreds of partners in its Trusted Access for Cyber program, with plans to expand to thousands in the coming weeks.
This is not the first time OpenAI has appeared to respond to Anthropic’s moves rather than set them. The company that led the public conversation around AI for years now seems increasingly reactive compared to the firm founded by its former employees.
Anthropic announced it had created Mythos, a cybersecurity-focused AI to which it granted access only to 40 organizations it considered “good actors,” such as Apple, Amazon, Microsoft, Google, and the Linux Foundation.
The reason both Anthropic and OpenAI have limited access to a select group of companies is that they claim these models are so powerful they can uncover vulnerabilities in systems that had never been discovered before, potentially putting networks at risk.
We are one of those people who try — sometimes successfully, sometimes not — to stay away from social media. To do that, we looked for alternatives, and one of the most effective has been Kinnu, a learning app that lets you explore topics of interest through daily lessons designed to be quick and easy to consume.
The platform positions itself as a kind of “best friend for your brain,” using learning techniques based on cognitive science to help improve information retention. With a colorful and intuitive design, you can learn about a wide range of subjects, from literature, mathematics, or art to social skills, history, or philosophy.
Once inside the app, you choose a subtopic to begin a learning path made up of short lessons. The system is gamified with streaks, daily challenges, smart sessions, and constant interaction, encouraging you to keep learning. Although there is a paid version, all the app’s main features are free.
Video uploaded by The Daily Show on April 17, 2026.
From the Nokia 1100 to the iPhone: The Five Cell Phones That Shaped the History…
The iPhone 17 Pro Max Surprises With An Unprecedented Image of the Moon’s Far Side Commander Reid Wiseman amazed viewers…
Google’s New Revolution Is Called “TurboQuant”, a Compression Algorithm A few days ago, Google unveiled…
Apple Confirms WWDC 2026: How to Watch It, Dates, and Possible Announcements The company led…
Why is Samsung Ending Sales of the Galaxy Z TriFold After Only 3 months? Samsung…
Apple Changes the Rules for Minors: Applies to iPhone, iPad, and Mac Users who do…