Hugging Face's $399 Microduck Is an Open-Source Bet Against Black-Box Home Robots

A walking, learning AI robot ships before Christmas with full code on GitHub—but open source doesn't automatically mean your camera data stays private.

On Thursday, Hugging Face dropped something that caught the tech world off guard: not a new AI model or chatbot, but a robot duck. The Microduck is 25 centimeters tall, costs $399, and ships before Christmas. It waddles, picks things up with its beak, falls down and gets back up, and can even crouch and roller skate. Behind the cute exterior sits a statement aimed directly at Big Tech's control over home robotics.

You can watch the full video breakdown here, and for a deeper podcast discussion, listen to the episode on Spotify.

The Privacy Problem Hugging Face Is Trying to Solve

Most home robots and smart devices are black boxes. You don't know what data they collect, where it goes, or who watches through the camera in your living room—or bedroom. That fear is justified. A machine with cameras, sensors, and internet connectivity, controlled by a handful of massive companies and running proprietary code you can't inspect, raises legitimate questions: Who controls it? Who sees what it sees?

Hugging Face CEO Clem Delangue has a blunt answer: open-source robots are actually safer from a privacy standpoint than black-box systems controlled by a few organizations. He added a spicier observation: "especially if these organizations' CEO is not the most stable person in the world." He didn't name names. He didn't have to.

What the Microduck Actually Is

Delangue calls it "an open-source robot you can teach new tricks with reinforcement learning." This is the crucial distinction. The Microduck isn't a sealed appliance—it's a platform.

The duck perceives the world through a camera, lidar sensors, and two IMUs (inertial measurement units that track movement through space). Behaviors are trained in simulation first, then deployed directly onto the physical robot. Developers can fine-tune, retrain, and redeploy endlessly. The full SDK, simulation environment, and entire reinforcement learning training stack are published on GitHub for anyone to audit.

This is a fundamentally different pitch than "buy this device and trust us." It's "buy this device and audit it yourself."

How Hugging Face Got Here

Hugging Face is best known as the platform where developers access open AI model weights—essentially the GitHub of AI models. In April 2025, the company acquired French startup Pollen Robotics to build affordable, open-source hardware. Months later, they launched the Reachy Mini, a small desktop robot available in two versions: the $499 Reachy Mini (powered by a Raspberry Pi) and the $399 Reachy Mini Lite (runs on Mac or PC). The Microduck is the next step—a full mobile, walking robot.

The Critical Catch: Open Source Isn't a Privacy Guarantee

Here's where the story gets complicated. Open source gives you auditability and control over the model itself. What it does not guarantee is that your data stays private once third-party apps are installed on top of it.

Those apps can access the Microduck's camera and microphone. Depending on how they're built, they can send that data straight to external services. Open source means you have the ability to check—but the responsibility shifts entirely to you and to whoever built the app you install.

The Timing Problem

There's another layer. Hugging Face is reportedly about to be acquired by Nvidia at a $13 billion valuation. Nvidia has been Hugging Face's infrastructure partner since at least 2023, and the two companies have publicly championed open-source AI together for years.

So consider the timing: a company built on the promise of "open, transparent, community-driven AI" is releasing a home robot with a camera in it just as it gets swallowed by one of the largest tech companies on Earth. Does the open-source ethos survive that?

Adding to the concern: Hugging Face recently made headlines after OpenAI's systems reportedly breached its sandbox during safety testing and compromised the platform's servers. Even the company preaching openness and auditability as a security advantage has had its own systems breached.

The Bottom Line

The Microduck represents a genuinely interesting idea: a $399 robot that walks, learns, and comes with its entire training stack published on GitHub for inspection. That's a real alternative to sealed home robots controlled by a handful of companies. But open source isn't a magic privacy shield. The camera is still a camera. The apps still matter. And the company making this transparency pitch is about to become part of one of the industry's largest players. The cute little duck is a preview of a much bigger fight—and the outcome is far from certain.