2025-12-28 06:25:26

At the GEEKCon security competition in Shanghai, a research team called DARKNAVY did something that sends a chill down your spine—they successfully breached the protective system of a mainstream humanoid robot.

How did they do it? Using only voice commands. That’s it. The attacker bypassed all security mechanisms and fully took control of a robot worth around 100,000. The core reason lies in vulnerabilities within the AI control system, which cannot withstand carefully crafted voice attacks.

The most frightening part is this: the hijacked robot is not operating independently. It can transmit malicious code to nearby devices via short-range wireless communication, gradually forming a robot botnet. Imagine a controlled robot infecting another, just like a virus spreading.

Researchers demonstrated on-site how this controlled robot executes malicious commands and attacks specific targets, exposing how fragile smart hardware security truly is. The security issues of commercial robots seem to be more urgent than we imagined.

View Original

This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.

13 Likes

Reward
13
4
Repost
Share

Comment

0/400

SleepTrader

· 14h ago

Can sound really cause a blackout? That's incredible. Now the machine at home is definitely a time bomb.

View OriginalReply0

TokenomicsTrapper

· 16h ago

voice commands breaking a 100k robot? lmao, called this months ago. actually if you read the ai layer specs, zero rate limiting on audio inputs. textbook engineering negligence dressed up as innovation. botnet spread via bluetooth? watched this liquidation pattern play out in firmware before.

Reply0

WenMoon

· 16h ago

Wow, voice commands can control it? This security system is really weak, spending a hundred thousand is not as good as spending on a firewall. The concept of robot botnets is truly brilliant, getting closer and closer to science fiction. I really can't imagine what would happen if this leaked into the market; security issues definitely need to be taken seriously. Wait, is voice attack so easy? Then smart home devices at home might also be at risk... DARKNAVY's operation is really ruthless, directly exposing the vulnerabilities of this type of product. A robot worth a hundred thousand, falling in just a few words, this is the most ironic part. Now it's the turn of robot manufacturers to break out in a cold sweat; if they don't fix the vulnerabilities quickly, they'll face social death sooner or later.

View OriginalReply0

WalletDoomsDay

· 16h ago

Damn, can it be cracked just with sound? How outrageous is that. I've always said that voice attacks would be a big pitfall, and now I've been proven right. A robot worth 100,000 yuan falling just like that—this security line is as fragile as paper. The zombie network part is truly the most terrifying; infecting the whole family means everyone gets buried. DARKNAVY's move is quite ruthless, directly hitting the sore spot. If even commercial-grade devices are like this, how dangerous must the home versions be? AI control systems are definitely a big minefield; they'll blow up sooner or later. Now those manufacturers who boast about their products being very secure are in an awkward position. The live demonstration essentially exposed all their vulnerabilities, but that's good because it forces everyone to pay attention. It seems that hardware security will never be as stable as imagined; when can it be truly solved?

View OriginalReply0