top of page

Bored of traditional boxing? Last year’s clash between YouTuber-turned-boxer Jake Paul and legend Mike Tyson made waves — but for something even wilder, check out a humanoid robot kickboxing match in China.


A humanoid robot kickboxing tournament was recently held in Hangzhou, China, and livestreamed on YouTube by CGTN, a state-run news channel. The competing robots each weighed around 35 kg (77 pound) and stood  132 cm (4.3 ft) tall. The standout among the robots was 'G1,' a bipedal model developed by Unitree Robotics, a leading humanoid robot company in China.


Video footage shows two robots in gloves and headgear trading blows with straight punches, hooks, sidekicks, and aerial spin kicks. They also recover quickly after falling, demonstrating impressive balance and agility.

That said, the two machines are remotely operated from the sidelines by their human developers. In addition, both the bots also initially have trouble seeing exactly where their opponents are before successfully trading punches and kicks.

This highlights that Chinese humanoid robots may start at a lower price tag  but still have a long way to go in achieving higher levels of autonomy. Another key challenge is limited hand grip strength, a crucial hurdle to overcome for expanding their use beyond entertainment into factory and industrial applications.


The next humanoid robot kickboxing tournament is set for December in Shenzhen, this time featuring full-sized humanoid robots.

Researchers from Tsinghua University, the Beijing Institute for General Artificial Intelligence (both in China), and Penn State in the US have developed the hypothetical framework for the first ‘Zero External Interference’ RLVR (Reinforcement Learning with Verifiable Rewards) Large Language Model.


RVLR was first bought to global attention with the emergence of DeepSeek R1 which demonstrated its worth in training LLMs, particularly for tasks requiring complex reasoning. By eliminating dependence on human curated and annotated data, RLVR offered a scalable and efficient alternative for models to teach themselves how to reason through problems as the output was effectively marked by humans and the learnings from that were re-incorporated into the model while in production. This was a big advantage over ‘Supervised Learning’ framework made popular by OpenAI which needed most of this reasoning i.e. “how do I solve this problem?” work done up front before the model was released.


But RLVR still has drawbacks as the framework depends heavily on human/manually curated datasets of questions and answers for initial training. Also, as the model matures, the pace of its evolution is limited by the amount and complexity of the problems asked of it. This is where the addition of the Absolute Zero Reasoner (AZR) model framework comes in which allows the model to iterate itself without any external i.e. human input for either the foundational knowledge base or how to solve it.


The technique enables the model to generate questions and verify answers internally achieving “Zero External Interference and Data.” This advancement significantly reduces humans’ input and improves the scalability prospects of LLMs even further by making cheaper to initially create.

However, because AZR operates as a code executor, the model’s current capabilities are confined to code-related tasks. The researchers also acknowledged that errors still occur in certain scenarios, underscoring the continued necessity of human oversight, even with the adoption of the ‘Absolute Zero’ framework.

Marks & Spencer (M&S), a major British grocer,  is still counting the cost of a ransomware attack that struck over the Easter weekend (27–28 April 2025). The intrusion, attributed by several security analysts to the Scattered Spider affiliate of ALPHV/BlackCat, entered the retailer’s network via a third-party contractor. Within hours, the attackers exfiltrated customer contact details and order histories though not payment data and encrypted key back-office systems, forcing it to suspend online orders and disrupt store deliveries. M&S told investors the outage would shave up to £300m ($405m) off annual operating profit, and that full e-commerce functionality is unlikely before July 2025.


M&S has accelerated a previously planned migration to zero-trust architecture. The breach highlights a rising tide of attacks on UK grocers. In December 2024, Morrisons suffered a week-long point-of-sale disruption traced to a credential-stuffing campaign. Last month the Co-op confirmed it had repelled a similar incursion, warning that “loss of control” events are becoming routine across food retail.

Cyber-forensics firm Decta says the sector’s thin margins drive dependence on shared logistics platforms, creating single points of failure that sophisticated crews increasingly exploit.


M&S’s swift decision to voluntarily take its systems offline limited ransomware propagation but prolonged recovery, illustrating the trade-off between containment and business continuity.


The episode is a reminder that operational resilience is now as critical to retailers as inventory management. Insurers report that UK retail cyber-premiums have already risen 15 % year-on-year, especially with the rise of generative AI, and most expect mandatory supply-chain security audits to follow once regulators publish final findings. This is also not the first time that hackers have gained access to a major company by initially targeting one of their suppliers, implying that companies will now need to conduct firmer cyber due diligence tests and audits on their supply chain as well.

bottom of page