Robot ambulances, robot protests, and China says the US is now copying them

Today in Humanoid Daily...

• JoyIn turns the distillation tables on OpenAI
• China's biggest angel round just hit a one-year-old robot startup
• Japan built an ambulance for broken humanoids
• Thirty robots protest for AI regulation in Warsaw
• And the number of the day: 3,600%

My name is John Koetsier, and this is Humanoid Daily for HumanoidMegaHub.com.

JoyIn flips the distillation playbook on OpenAI

For years, U.S. AI companies have accused Chinese firms of distillation. That's taking the training outputs from frontier models and using them to improve cheaper domestic alternatives.

It's kind of cheating: a way to build serious capability without spending billions in compute. 

It's became a whole geopolitical thing: Anthropic complained about it, the U.S. cybersecurity agency flagged it and six Chinese companies were implicated, including DeepSeek and Alibaba.

Now a Chinese humanoid startup is pulling an Uno Reverse card.

JoyIn, an Ant Group-backed robotics company based in Suzhou, published an open letter to OpenAI this week accusing the U.S. AI giant of lifting concepts directly from JoyIn's own research. CEO Guo Renjie says JoyIn presented its "extraterrestrial visitor" AI framework in Silicon Valley several weeks ago and then, on September 6, OpenAI's Chief Scientist published a paper called "An Alien Mind." 

Similar title, similar outer-space inspiration, suspicious timing. 

Guo also points to overlapping technical approaches, including recursive self-improvement and AI-optimized compute, plus what he describes as obvious aesthetic similarities in both companies' model web pages.

CNBC, which broke this story, notes it could not independently verify the claims. Some of the referenced concepts are broad categories in AI research that many teams work on simultaneously. OpenAI didn't respond to a request for comment. And yes, Guo has started the process of filing a lawsuit.

Meanwhile, separate from the legal drama, JoyIn simultaneously released its Aether robotics model, which the company says takes a perceptive rather than text-based approach to controlling humanoids. Aether cuts training time by two-thirds, the company says, and achieves a 90% first-attempt task completion rate ... which is very impressive. 

(Interesting: the lead developer previously worked at Figure, the U.S. humanoid company.)

Will the lawsuit go anywhere? The allegations are probably not easy to prove. But the optics, at any rate, matter.

A one-year-old Chinese startup just raised a $74 million angel round

$74 million angel round?

$74 million angel round!

$74 million angel round.

Umm ... wow. Kinetix AI, founded in September 2025, has closed an angel funding round of over 500 million yuan, roughly $74.5 million. The round was led by Vertex Ventures, the VC arm of Singapore's Temasek Holdings, with F&G Venture and Wanshi Capital also participating.

That is perhaps the largest "angel" round I've ever heard of. It's larger than most company's series C or D rounds. Impressive!

Kinetix AI's robot is KAIBot: 1.73 meters tall, 65 to 70 kilograms, and 115 degrees of freedom. For context, the human body has more than 200 degrees of freedom. 115 is a lot for a machine, perhaps the most I've heard of.

Also, the founding team is notable. CEO Yu Jie was the general manager of Huawei's commercial vehicle autonomous driving unit, reaching level 22 internally and overseeing Huawei's L4 autonomous mining trucks. Co-founder Zheng Cunyuan was previously head of XPeng's humanoid robot hardware division. Additional co-founder Luo Ping is associate dean for research at the University of Hong Kong's School of Computing and Data Science.

Kinetix is betting on what they call the "highly humanoid" path: robots built close to human proportions, designed to leverage the massive amount of human motion data that already exists in the world.

This has risks, of course. It costs more, takes longer, and has a more complex supply chain. The competing approaches offer cheaper hardware and faster deployment. But, get it right, and you've got a (instant?) winner.

Interestingly, at the 2026 World Humanoid Robot Games, a Kinetix-equipped robot played exhibition table tennis against world champion Ding Ning.

Japan built an ambulance for broken humanoid robots

What happens when a humanoid robot breaks down mid-shift?

You call the robot ambulance, apparently.

(Today is probably the first time you've ever heard "robot ambulance," I'm guessing. It is for me too.)

GMO Air, a subsidiary of Japanese IT giant GMO Internet Group, just unveiled a van designed specifically for humanoid robot maintenance. The vehicle carries engineers, repair equipment, spare parts, and a replacement humanoid robot ready to swap in immediately when one goes down on site. 

It's got purple roof lights, ambulance-style markings, mobile repair workshop inside.

If the broken robot can't be fixed on the spot, it rides back to the repair base while the replacement takes over the shift. GMO says it came up with the idea after experiencing real disruptions from damaged humanoids being sent away for repair, leaving jobs uncovered.

Right now there's one vehicle, stationed at GMO's Tokyo headquarters. The company says it could add more as demand grows.

This is a small story with an interesting signal: robots are getting deployed in actual workplaces now. Actual workplaces need actual support infrastructure. 

And an ambulance for humanoids is exactly the kind of mundane, logical service that shows up when robots are genuinely working, not just doing demos.

Thirty robots rallied outside Poland's digital affairs ministry

About 30 robots turned up outside Poland's digital affairs ministry in Warsaw this week, holding signs and flags, chanting slogans, demanding stronger AI regulation. The advocacy group Democratism organized the protest to push debate about AI's impact on jobs and the labor market.

Poland's Digital Affairs Minister Krzysztof Gawkowski came out and met with them.

It's performance art, obviously. 

Robots don't have demands (yet). Humans with robots have demands. 

But performance art in front of an actual minister is also politics, and the image of humanoid robots literally protesting for regulation of themselves is the kind of thing that gets shared, clipped, and remembered. In the EU, where AI regulation is actively being written, that matters.

The number of the day: 3,600%

That's how much Apptronik's shares on private secondary markets have surged, according to Benzinga.

A few important caveats before you go update your net worth spreadsheet. Private secondary market shares are not public stock. They're highly illiquid instruments traded between insiders and early investors in opaque markets, extremely sensitive to momentum and hype. 

A 3,600% surge doesn't mean revenue went up 36x.

What it does mean is that private investors are extraordinarily bullish on humanoid robots right now, and Apptronik, which holds a real NASA partnership and is actively deploying Apollo robots in factory environments, is a prime beneficiary of that enthusiasm.

It also means the private market has already priced in an enormous amount of future success for a company that hasn't yet shipped at scale. That's a great problem to have. It's also a serious one. Expectations at that level leave almost no room for the normal, inevitable stumbles of building a new category of machine.

Sources

• CNBC — A Chinese humanoid startup flips 'distillation' claim on OpenAI as it releases a new robotics model
• BigGo Finance — China's Embodied AI Startup Kinetix AI Raises Over 500 Million Yuan in Angel+ Round
• Euronews — Japanese company unveils 'robot ambulance' for on-site humanoid repairs
• South China Morning Post — Robots protest to demand improved AI regulation
• Benzinga — Apptronik's Private Market Share Soars 3,600% on Humanoid Robot Hype