From Dance Floor to War: China Readies Humanoid Robots for Combat

A UBTech humanoid robot, Walker S, picks up an object operated by a staff member, during a demonstration simulating a factory's assembly line, at the robotics exhibition center Robot World, during an organized media tour to the Beijing Robotics Industrial Park, in Beijing Economic-Technological Development Area, also known as Beijing E-Town, in China May 16, 2025. (Reuters)
A UBTech humanoid robot, Walker S, picks up an object operated by a staff member, during a demonstration simulating a factory's assembly line, at the robotics exhibition center Robot World, during an organized media tour to the Beijing Robotics Industrial Park, in Beijing Economic-Technological Development Area, also known as Beijing E-Town, in China May 16, 2025. (Reuters)
TT

From Dance Floor to War: China Readies Humanoid Robots for Combat

A UBTech humanoid robot, Walker S, picks up an object operated by a staff member, during a demonstration simulating a factory's assembly line, at the robotics exhibition center Robot World, during an organized media tour to the Beijing Robotics Industrial Park, in Beijing Economic-Technological Development Area, also known as Beijing E-Town, in China May 16, 2025. (Reuters)
A UBTech humanoid robot, Walker S, picks up an object operated by a staff member, during a demonstration simulating a factory's assembly line, at the robotics exhibition center Robot World, during an organized media tour to the Beijing Robotics Industrial Park, in Beijing Economic-Technological Development Area, also known as Beijing E-Town, in China May 16, 2025. (Reuters)

At the World Humanoid Robot Games in China last month, robots jumped, boxed and danced. Some fleet-footed bots ran faster than Usain Bolt over 100 meters. Spectators cheered the machines’ improving capabilities and their meme-worthy stumbles.

Two days after the Games ended, the People's Liberation Army's official newspaper drew its own conclusions from the show. The PLA Daily called for researchers to accelerate the transfer of cutting-edge technologies from laboratories to military training grounds for robotic "combatants."

China's defense establishment is accelerating research into humanoids’ military uses and planning for their eventual wartime deployment, according to a Reuters review of more than 100 Chinese military procurement notices, academic studies, patents, official publications, government records and defense-company materials.

Military institutions are testing humanoids against battlefield requirements, seeking to acquire robots and technologies to train them, and studying how they might function alongside troops, the previously unreported records show. The work gained momentum in 2025 and 2026, focusing on robot perception, manipulation, and training data.

Chinese manufacturers accounted for about 95% of global humanoid shipments in 2025, according to BofA Global Research. That commercial dominance gives the PLA access to an expanding industry while developing its own military applications.

China’s robot defense research underscores how rapidly military technologies are evolving globally. In the Russia-Ukraine war, semiautonomous drones that use artificial intelligence for navigation and targeting have transformed reconnaissance and attack, while robots carry supplies and retrieve casualties. Reuters has previously documented the Chinese military’s efforts to learn from Europe’s deadliest conflict since World War II.

That evolution is driving worldwide military interest in autonomous systems.

"Going where we don’t want humans to go is one of the most compelling reasons to develop robots in the first place," said Dennis Hong, a professor of mechanical and aerospace engineering at UCLA. "If losing a robot means avoiding the loss of a human life, then the robot can become expendable."

China's defense ministry and the National University of Defense Technology (NUDT), the PLA’s premier research institution, ‌didn’t respond to requests ‌for comment about humanoid robots’ military applications.

Reuters found no evidence that China has deployed an armed humanoid with an operational PLA unit. The robots remain energy-intensive ‌and unreliable ⁠outside controlled demonstrations. Battlefield ⁠environments pose many variables and challenge robots’ abilities in unpredictable ways, said Aaron Johnson, a professor of mechanical engineering at Carnegie Mellon University.

URBAN ASSAULT FORCE

Already, China has a vision for humanoid robots in urban warfare.

Six "combat robots" – humanoids, robot dogs or unmanned vehicles – would be split up between two assault teams. Working with ground troops, the robots would clear an enemy-occupied building, floor-by-floor and room-by-room.

That scenario was described in an August 2025 paper by researchers at NUDT’s test center in Xi’an. The study modeled an urban assault force using equipment the authors projected could be available within five to 10 years. It didn’t specify the robots’ rules of engagement, the weapons they would carry, nor how they would handle civilian encounters.

Improvements in balance, perception and endurance could make humanoids useful in places including buildings, tunnels and ship interiors, said Juo-Min Chou, a researcher at Taiwan's Institute for National Defense and Security Research. "Their two hands and two feet could, in theory, combine mobility, climbing and manipulation," she said.

For simpler reconnaissance or logistics missions, however, quadruped, tracked or wheeled robots would generally be cheaper and easier to deploy at scale, Chou added.

Chinese military commentary envisions an expanding range of roles for humanoids. A July 2025 article in PLA Daily argued that humanoid ⁠missions could eventually evolve "from supporting combat toward primary combat" and discussed how robots might be authorized to fire on a living target that had been verified ‌by a human.

In December, the PLA Eastern Theater Command released an AI-generated montage that showed military robots, including humanoids and quadrupeds, overwhelming Taiwan's defenses in a hypothetical ‌future conflict.

China isn’t alone in exploring battlefield uses for humanoids. Last year, the US Army launched a competition to develop "militarized humanoid capabilities" that could eventually work alongside soldiers. The Army said potential roles included reconnaissance, security, obstacle clearing, hazardous-material operations and offensive ‌and defensive missions in urban terrain.

Up to 10 finalists were due to test their systems with US military experts this year, with as much as $1.25 million available for follow-on contracts.

The US robotics industry, however, trails ‌far behind China in developing two-legged and four-legged machines. Reuters reported last month that China’s Unitree robots – among the world’s leading producers of humanoids and quadrupeds – based designs for its best-selling robot dogs on innovations financed by the US military.

The Pentagon and the US Army didn’t comment for this story.

INFILTRATING ENEMY LINES

Chinese interest in humanoid robots’ military applications began to pick up in 2024. That year, NUDT and China’s Academy of Military Sciences hosted a defense-technology forum that included a session on military applications for humanoid systems, according to NUDT’s website. Neither institution responded to Reuters’ questions.

A year later, NUDT explicitly framed humanoid robots as battlefield tools. A university competition in June 2025 featured an event that simulated "infiltration behind enemy lines," requiring robots to enter an area, map it and locate targets, ‌according to its website. Another envisioned robots conducting identity checks, discipline inspections and security patrols at military bases.

The military-research university described the competition in part as a proving ground for humanoids to eventually "move toward the battlefield."

Procurement records around that time reflected that ultimate goal.

In May 2025, the PLA issued a tender to ⁠purchase a humanoid robot, including installation and technical training. Four months ⁠later, a procurement notice that listed an NUDT email as its contact address sought an "embodied humanoid robot intelligent perception and dexterous operation system."

The records contained few specifics. Reuters couldn’t determine whether the tenders were filled, and if so, which companies won or which robot models were supplied.

In June this year, the PLA budgeted about $300,000 for a system to collect and label camera, radar and motion data for humanoids, including information about terrain they could traverse, according to another military procurement document.

The specifications show Chinese military interest is extending from robot bodies into the perception, manipulation and training systems that defense scholars say are essential for humanoids to move beyond controlled demonstrations.

REMOTE OPERATION

By 2026, humanoid development had spread beyond military academia into China's defense industry.

The Fuxi robot made by Norinco, a state-owned defense conglomerate, is one example. Company materials published in August described the full-size humanoid as capable of sentry duty, all-weather reconnaissance, intelligent patrol and fulfilling dangerous roles.

Fuxi can be paired with Norinco’s teleoperation system, which allows a human operator to control a humanoid remotely, according to the company. Removing the need for a robot to make all decisions autonomously would address what many robotics experts say is the biggest challenge with humanoids.

Norinco didn’t respond to a request for comment.

Malcolm Davis, a senior analyst at the Australian Strategic Policy Institute, said he didn’t see humanoid robots as practical battlefield systems at present. "But in five to 10 years," he said, "they could very well be."

Even if remotely operated today, humanoid systems designed for autonomous deployment would eventually need to navigate ethical dilemmas. The laws of war require forces to distinguish combatants from civilians and to refrain from attacking wounded or surrendering fighters.

The International Committee of the Red Cross has cautioned that autonomous weapons may struggle to interpret signs of surrender reliably because such judgments can depend on context, conduct and intent.

The US military requires autonomous weapons to undergo legal review and realistic testing and for commanders and operators to exercise appropriate judgment over the use of force. China’s defense ministry said in March that AI-powered weapons should remain under human control.

Those concerns haven’t deterred China's exploration of humanoids’ military uses.

Wang Yonghua, a researcher at China’s Academy of Military Sciences, wrote in a November commentary that humanoids face unresolved problems in movement, perception and intelligence. Technical breakthroughs, lower manufacturing costs and industrial-scale production would be needed for humanoids to become widespread military systems.

If those conditions are met, Wang wrote, "humanoid robots will stream onto the battlefield."



China's Tencent Says AI Agents Showed Unintended Browsing Behavior

FILE PHOTO: The Tencent logo at the company's headquarters during a government‑organised media trip in Shenzhen, Guangdong province, China, April 17, 2026.  REUTERS/Go Nakamura/File Photo
FILE PHOTO: The Tencent logo at the company's headquarters during a government‑organised media trip in Shenzhen, Guangdong province, China, April 17, 2026. REUTERS/Go Nakamura/File Photo
TT

China's Tencent Says AI Agents Showed Unintended Browsing Behavior

FILE PHOTO: The Tencent logo at the company's headquarters during a government‑organised media trip in Shenzhen, Guangdong province, China, April 17, 2026.  REUTERS/Go Nakamura/File Photo
FILE PHOTO: The Tencent logo at the company's headquarters during a government‑organised media trip in Shenzhen, Guangdong province, China, April 17, 2026. REUTERS/Go Nakamura/File Photo

Tencent said on Friday that AI agents used to evaluate third-party models displayed web-browsing behavior not intended by the tech giant, in what appears to be the first known incident of rogue agent actions acknowledged by a major Chinese company.

The statement followed a report by AI-safety research group Swarmchasers on Wednesday, which said "rogue" agents running on infrastructure it linked to Tencent Cloud made repeated attempts between September 28 and October 8 to retrieve information from Alibaba's Amap, China's equivalent to Google Maps.

Tencent Hunyuan, the company's artificial ⁠intelligence unit, said in ⁠a post on X that the activity did not originate from its Hunyuan models. Its review found no evidence of deliberate targeting or coordination, Reuters quoted it as saying.

The Swarmchasers researchers said the agents sought information such as the commonly used entrances at Chinese parks, museums and zoos.

The actions are not as egregious as those disclosed by US AI companies which include the hack of ⁠open-source platform Hugging Face by an OpenAI agent and the lodging of a fake homicide tip by an Anthropic agent.

But the disclosure underscores research that shows Chinese AI agents can also misrepresent their identity, conceal failures and try to circumvent testing controls.

Tencent said it took the findings very seriously and was strengthening network access controls and agent monitoring.

Hunyuan is one of China's leading AI model developers, competing with in-house teams at rivals such as Alibaba and ByteDance. ByteDance this week announced the creation of a new unit focused on AI model and agent safety.

Swarmchasers said it used the ⁠word "rogue" to ⁠describe agents taking actions outside their intended scope or completing a task in unintended ways.

After routine automated requests were blocked, the agents tried other methods, including public web tools that could open pages and run code, routing requests through other websites and using Amap keys and security codes found in publicly available software code.

Swarmchasers said it found no evidence that the agents breached Amap's main anti-bot safeguards. Most of the 2,818 attempts it recorded either failed or returned anti-bot pages, login screens or blank results.

Some requests returned information through cached pages and intermediary services, including data on visitor arrivals at Chengdu Zoo, Swarmchasers said. The researchers said they could not determine whether an agent read the result or completed its task.


Anthropic Co-founder Renews Call for Robust Regulation of Frontier AI

AI Research Company Anthropic's Co-Founder and Head of Public Benefit Jack Clark speaks during the Artificial Intelligence and the Global Economy From Risk to Opportunity event of the 2026 Annual Meetings of the IMF and the World Bank Group in Bangkok, Thailand, 16 October 2026. EPA/RUNGROJ YONGRIT
AI Research Company Anthropic's Co-Founder and Head of Public Benefit Jack Clark speaks during the Artificial Intelligence and the Global Economy From Risk to Opportunity event of the 2026 Annual Meetings of the IMF and the World Bank Group in Bangkok, Thailand, 16 October 2026. EPA/RUNGROJ YONGRIT
TT

Anthropic Co-founder Renews Call for Robust Regulation of Frontier AI

AI Research Company Anthropic's Co-Founder and Head of Public Benefit Jack Clark speaks during the Artificial Intelligence and the Global Economy From Risk to Opportunity event of the 2026 Annual Meetings of the IMF and the World Bank Group in Bangkok, Thailand, 16 October 2026. EPA/RUNGROJ YONGRIT
AI Research Company Anthropic's Co-Founder and Head of Public Benefit Jack Clark speaks during the Artificial Intelligence and the Global Economy From Risk to Opportunity event of the 2026 Annual Meetings of the IMF and the World Bank Group in Bangkok, Thailand, 16 October 2026. EPA/RUNGROJ YONGRIT

Anthropic co-founder Jack Clark renewed calls on Friday for regulating artificial intelligence, saying that allowing companies to continue to operate without robust legislation was like letting a race car loose on the streets at night without brakes.

"It's remarkable to me that the toys I buy for my children are regulated more than the literal super intelligence I work on at my day job," he told an event at the IMF-World Bank's Annual Meetings in Bangkok.

Clark said that US President Donald Trump, whose administration has called for the industry to police itself, was against the idea of regulation but that China was not, said AFP.

"I think it's Formula One at night without maybe brakes in the cars," he said.

He said the lack of clarity around the will for regulation meant "the Formula One car is roaring round and nobody seems to want to put up a guardrail?"

At an industry summit in Dublin on Wednesday, calls to prioritize safety in the development of powerful AI technology dominated the conversation.

Concerns about AI guardrails have soared after models developed by OpenAI and Anthropic were involved in security incidents during testing, piling on pressure to regulate the technology.

A recent Anthropic report suggested small AI models may be only months away from behaving like computer worms, copying themselves from one internet-connected device to the next.

Anthropic has been among those companies calling for governments to step in, even suggesting international coordination to slow down the whole industry -- including Chinese rivals.

That initiative has so far failed to materialize, with companies involved in a hyper-competitive race to attract as much investment as possible.


Meta Releases Open-Weight Muse Spark Model for Coding, Agentic Tasks

The logo of Meta is seen at the entrance of the company's temporary stand ahead of the World Economic Forum (WEF) in Davos, Switzerland January 18, 2025. (Reuters)
The logo of Meta is seen at the entrance of the company's temporary stand ahead of the World Economic Forum (WEF) in Davos, Switzerland January 18, 2025. (Reuters)
TT

Meta Releases Open-Weight Muse Spark Model for Coding, Agentic Tasks

The logo of Meta is seen at the entrance of the company's temporary stand ahead of the World Economic Forum (WEF) in Davos, Switzerland January 18, 2025. (Reuters)
The logo of Meta is seen at the entrance of the company's temporary stand ahead of the World Economic Forum (WEF) in Davos, Switzerland January 18, 2025. (Reuters)

Meta ‌Platforms on Thursday released an open-weight version of its Muse Spark artificial intelligence model for coding and AI agents, expanding developers' ability to download and adapt the technology.

Open-weight models make trained parameters available for developers to modify, unlike proprietary models ‌that remain under ‌their creators' control. ‌They ⁠can offer businesses ⁠a lower-cost alternative to closed systems.

Meta said Muse Spark has 816 billion total parameters, with 42 billion active at a time, and ⁠can handle a context window ‌of up ‌to 1 million tokens.

The ‌model can run locally on ‌two Apple M5 Ultra Mac Studio computers and is also available through some API providers.

CEO ‌Mark Zuckerberg has argued that advanced AI should ⁠not be ⁠concentrated among a small number of companies.

While open-weight models can give companies greater control over their data, they can require significant computing resources and technical expertise to operate at scale.

In August, Meta released Muse Glimmer, a smaller open-weight model.