Nvidia refutes chip backdoor allegations as China launches probe

Nvidia has firmly denied claims that its AI chips contain backdoors allowing remote control or tracking, following questioning by China’s top cybersecurity agency.

The investigation, which focuses on the H20 chip explicitly designed for the Chinese market, comes as Beijing intensifies scrutiny over foreign tech used in sensitive systems.

The H20 was initially blocked from export in April under US restrictions, but is now expected to return to Chinese shelves.

China’s Cyberspace Administration (CAC) summoned Nvidia officials to explain whether the chip enables unauthorised access or surveillance. The agency cited demands from US lawmakers for mandatory tracking features in advanced AI hardware as grounds for its concern.

In a statement, Nvidia insisted it does not include remote access capabilities in its products, reaffirming its commitment to cybersecurity.

Meanwhile, China’s state-backed People’s Daily questioned the company’s trustworthiness, stating that ‘network security is as vital as national territory’ and warning against reliance on what it described as ‘sick chips’.

The situation highlights Nvidia’s delicate position as it attempts to maintain dominance in China’s AI chip market while complying with mounting US export rules.

Tensions have escalated since similar actions were taken against other US firms, including a 2022 ban on Micron’s chips and recent antitrust scrutiny over Nvidia’s Mellanox acquisition.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

DW Weekly #223 – AI race heats: The US AI Action Plan, China’s push for a global AI cooperation organisation, and the EU’s regulatory response

 Logo, Text

25 July – 1 August 2025


 Book, Comics, Publication, Adult, Male, Man, Person, Face, Head, Clothing, Coat, Hat, James Montgomery Flagg

Dear readers,

Over the past week, the White House has launched a sweeping AI initiative through its new publication Winning the Race: America’s AI Action Plan, an ambitious strategy to dominate global AI leadership by promoting open-source technology and streamlining regulatory frameworks. America’s ‘open-source gambit’, analysed in detail by Dr Jovan Kurbalija in Diplo’s blog, signals a significant shift in digital policy, intending to democratise AI innovation to outpace competitors, particularly China.

Supporting this bold direction, major tech giants have endorsed President Trump’s AI deregulation plans, despite widespread public concerns regarding potential societal impacts. Trump’s policies notably include an explicit push for ‘anti-woke’ AI frameworks within US government contracts, raising contentious debates about the ideological neutrality and ethical implications of AI systems in governance.

In parallel, China has responded with its own global AI governance plan, proposing the establishment of an international AI cooperation organisation to enhance worldwide coordination and standard-setting. Thus, it is not hard to conclude that there is an escalating AI governance competition between the two technological superpowers, each advocating distinctly different visions for the future of global AI development.

On the multilateral stage, the UN’s Economic and Social Council (ECOSOC) adopted a resolution: ‘Assessment of the progress made in the implementation of and follow-up to the outcomes of the World Summit on the Information Society’, through the Commission on Science and Technology for Development (CSTD), reaffirming commitments to implement the outcomes of the World Summit on the Information Society (WSIS).

Corporate strategies have also reflected these geopolitical undercurrents. Samsung Electronics has announced a landmark $16.5 billion chip manufacturing deal with Tesla, generating optimism about Samsung’s capability to revive its semiconductor foundry business. Yet, execution risks remain substantial, prompting Samsung’s Chairman Jay Y. Lee to promptly travel to Washington to solidify bilateral trade relations and secure the company’s position amid potential trade tensions.

Similarly, Nvidia has placed a strategic order for 300,000 chipsets from Taiwanese giant TSMC, driven by robust Chinese demand and shifting US trade policies.

Meanwhile, the EU has intensified regulatory scrutiny, accusing e-commerce platform Temu of failing mandatory Digital Services Act (DSA) checks, citing serious risks related to counterfeit and unsafe goods.

In the USA, similar scrutiny arose as Senator Maggie Hassan urged Elon Musk to take decisive action against Southeast Asian criminal groups using Starlink services to defraud American citizens.

Finally, the EU’s landmark AI Act commenced its implementation phase this week, despite considerable pushback from tech firms concerned about regulatory compliance burdens.

Diplo Blog – The open-source gambit: How America plans to outpace AI rivals by democratising tech

On 23 July, the US unveiled an AI Action Plan featuring 103 recommendations focused on winning the AI race against China. Key themes include promoting open-source AI to establish global standards, reducing regulations to support tech firms, and emphasising national security. The plan addresses labour displacement, AI biases, and cybersecurity threats, advocating for reskilling workers and maintaining tech leadership through private sector flexibility. Additionally, it aims to align US allies within an AI framework while expressing scepticism toward multilateral regulations. Overall, the plan positions open-source AI as a strategic asset amid geopolitical competition. Read the full blog!

For the main updates, reflections and events, consult the RADAR, the READING CORNER and the UPCOMING EVENTS section below.

Join us as we connect the dots, from daily updates to main weekly developments, to bring you a clear, engaging monthly snapshot of worldwide digital trends.

DW Team


RADAR

Highlights from the week of 25 July – 1 August 2025

registration 3938434 1280

But worries rise as many free VPNs exploit users or carry hidden malware

australia flag is depicted on the screen with the program code

From December, YouTube must block accounts for Australians under 16 or face massive fines.

aeroflot cyberattack silent crow belarus cyber partisans ukraine conflict

Belarusian and Ukrainian hackers claim responsibility for strategic cyber sabotage of Aeroflot.

2025%2F07%2Fbeautiful silhouette port machinery sunset

A NATO policy brief warns that civilian ports across Europe face increasing cyber threats from state-linked actors and calls for updated maritime strategies to strengthen cybersecurity and civil–military coordination.

whatsapp Italy Meta antitrust EU AGCM

AGCM says Meta may have harmed competition by embedding AI features into WhatsApp.

eu and google

The EU AI Code could add €1.4 trillion to Europe’s economy, Google says.

0a560533 72a4 4acc a3a6 84f2162044df

Tether and Circle dominate the fiat-backed stablecoin market, now valued at over $227 billion combined.

microsoft logo png seeklogo 168319

Brussels updates Microsoft terms to curb risky data transfers

66b3be5a4f47d87bdc945227 image1 min

AI use in schools is weakening the connection between students and teachers by permitting students to bypass genuine effort through shortcuts.

AI Depression Design MT 1200x900 1

Use of AI surveillance, including monitoring software, intensifies burnout, micromanagement feelings, and disengagement.

cybersecurity risks of generative ai

A majority of Fortune 500 companies now mention AI in their annual reports as a risk factor instead of citing its benefits.

man using laptop night workplace top view

The platforms lost more than $3.1 billion in the first half of 2025, with AI-powered hacks and phishing scams leading the surge.

US AI jobs Brookings Lightcast survey

AI jobs now span marketing, finance, and HR—not just tech.

jojickajoja27 quantum computing 91572e2a e7d0 40d4 b6e5 12dc1bca48c6 11zon

Google and Microsoft lead investment in advanced AI and quantum infrastructure.


READING CORNER
BLOG featured image 2025 The open source gambit

On 23 July, the US unveiled an AI Action Plan featuring 103 recommendations focused on winning the AI race against China. Key themes include promoting open-source AI to establish global standards, reducing regulations to support tech firms, and emphasising national security.

ChatGPT Image Jul 28 2025 at 10 13 23 PM

Tracking technologies shape our online experience in often invisible ways, yet profoundly impactful, raising important questions about transparency, control, and accountability in the digital age.

China says the US used a Microsoft server vulnerability to launch cyberattacks

China has accused the US of exploiting long-known vulnerabilities in Microsoft Exchange servers to launch cyberattacks on its defence sector, escalating tensions in the ongoing digital arms race between the two superpowers.

In a statement released on Friday, the Cyber Security Association of China claimed that US hackers compromised servers belonging to a significant Chinese military contractor, allegedly maintaining access for nearly a year.

The group did not disclose the name of the affected company.

The accusation is a sharp counterpunch to long-standing US claims that Beijing has orchestrated repeated cyber intrusions using the same Microsoft software. In 2021, Microsoft attributed a wide-scale hack affecting tens of thousands of Exchange servers to Chinese threat actors.

Two years later, another incident compromised the email accounts of senior US officials, prompting a federal review that criticised Microsoft for what it called a ‘cascade of security failures.’

Microsoft, based in Redmond, Washington, has recently disclosed additional intrusions by China-backed groups, including attacks exploiting flaws in its SharePoint platform.

Jon Clay of Trend Micro commented on the tit-for-tat cyber blame game: ‘Every nation carries out offensive cybersecurity operations. Given the latest SharePoint disclosure, this may be China’s way of retaliating publicly.’

Cybersecurity researchers note that Beijing has recently increased its use of public attribution as a geopolitical tactic. Ben Read of Wiz.io pointed out that China now uses cyber accusations to pressure Taiwan and shape global narratives around cybersecurity.

In April, China accused US National Security Agency (NSA) employees of hacking into the Asian Winter Games in Harbin, targeting personal data of athletes and organisers.

While the US frequently names alleged Chinese hackers and pursues legal action against them, China has historically avoided levelling public allegations against American intelligence agencies, until now.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

China’s Silk Typhoon hackers filed patents for advanced spyware tools

A Chinese state-backed hacking group known as Silk Typhoon has filed more than ten patents for intrusive cyberespionage tools, shedding light on its operations’ vast scope and sophistication.

These patents, registered by firms linked to China’s Ministry of State Security, detail covert data collection software far exceeding the group’s previously known attack methods.

The revelations surfaced following a July 2025 US Department of Justice indictment against two alleged members of Silk Typhoon, Xu Zewei and Zhang Yu.

Both are associated with companies tied to the Shanghai State Security Bureau and connected to the Hafnium group, which Microsoft rebranded as Silk Typhoon in 2022.

Instead of targeting only Windows environments, the patent filings reveal a sweeping set of surveillance tools designed for Apple devices, routers, mobile phones, and even smart home appliances.

Submissions include software for bypassing FileVault encryption, extracting remote cellphone data, decrypting hard drives, and analysing smart devices. Analysts from SentinelLabs suggest these filings offer an unprecedented glimpse into the architecture of China’s cyberwarfare ecosystem.

Silk Typhoon gained global attention in 2021 with its Microsoft Exchange ProxyLogon campaign, which prompted a rare coordinated condemnation by the US, UK, and EU. The newly revealed capabilities show the group’s operations are far more advanced and diversified than previously believed.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

China demands Nvidia explain security flaws in H20 chips

China’s top internet regulator has summoned Nvidia to explain alleged security concerns linked to its H20 computing chips.

The Cyberspace Administration of China stated that the chips, which are sold domestically, may contain backdoor vulnerabilities that could pose risks to users and systems.

Instead of ignoring the issue, Nvidia has been asked to submit technical documents and provide a formal response addressing these potential flaws.

The chips are part of Nvidia’s tailored product line for the Chinese market following US export restrictions on advanced AI processors.

The investigation signals tighter scrutiny from Chinese authorities on foreign technology amid ongoing geopolitical tensions and a global race for semiconductor dominance.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

China issues action plan for global AI governance and proposes global AI cooperation organisation

At the 2025 World AI Conference in Shanghai, Chinese Premier Li Qiang urged the international community to prioritise joint efforts in governing AI, making reference to a need to establish a global framework and set of rules widely accepted by the global community. He unveiled a proposal by the Chinese government to create a global AI cooperation organisation to foster international collaboration, innovation, and inclusivity in AI across nations.

China attaches great importance to global AI governance, and has been actively promoting multilateral and bilateral cooperation with a willingness to offer more Chinese solutions‘.

An Action Plan for AI Global Governance was also presented at the conference. The plan outlines, in its introduction, a call for ‘all stakeholders to take concrete and effective actions based on the principles of serving the public good, respecting sovereignty, development orientation, safety and controllability, equity and inclusiveness, and openness and cooperation, to jointly advance the global development and governance of AI’.

The document includes 13 points related to key areas of international AI cooperation, including promoting inclusive infrastructure development, fostering open innovation ecosystems, ensuring high-quality data supply, and advancing sustainability through green AI practices. It also calls for consensus-building around technical standards, advancing international cooperation on AI safety governance, and supporting countries – especially those in the Global South – in ‘developing AI technologies and services suited to their national conditions’.

Notably, the plan indicates China’s support for multilateralism when it comes to the governance of AI, calling for an active implementation of commitments made by UN member states in the Pact for the Future and the Global Digital Compact, and expressing support for the establishment of the International AI Scientific Panel and a Global Dialogue on AI Governance (whose terms of reference are currently negotiated by UN member states in New York).

Would you like to learn more about AI, tech and digital diplomacyIf so, ask our Diplo chatbot!

UBTech’s Walker S2 marks a leap towards uninterrupted robotic work

The paradigm of robotic autonomy is undergoing a profound transformation with the advent of UBTech’s new humanoid, the Walker S2. Traditionally, robots have been tethered to human assistance for power, requiring manual plugging in or lengthy recharges.

UBTech, a pioneering robotics company, is now dismantling these limitations with a groundbreaking feature in the Walker S2: the ability to swap its battery autonomously. The innovation promises to reshape the landscape of factory work and potentially many other industries, enabling near-continuous, 24/7 operation without human intervention.

The core of this advancement lies in the Walker S2’s sophisticated self-charging mechanism. When a battery begins to deplete, the robot does not power down. Instead, it intelligently navigates to a strategically placed battery swap station.

Once positioned, the robot executes a precise sequence of movements: it twists its torso, deploys built-in tools on its arms to unfasten and remove the drained battery from its back cavity, places it into an empty bay on the swap station, and then expertly retrieves a fresh, fully charged module.

The new battery is then securely plugged into one of its dual battery bays. The process is remarkably swift, taking approximately three minutes, allowing the robot to return to its tasks almost immediately.

The hot-swappable system mirrors the convenience of advanced electric vehicle technology, but its application to humanoid robotics unlocks unprecedented operational efficiency. Standing at 5 feet, 3 inches (approximately 160 cm) tall and weighing 95 pounds (about 43 kg), the Walker S2 is designed to integrate seamlessly into environments built for humans.

It has two 48-volt lithium batteries, ensuring a continuous power supply during the brief swapping procedure. While one battery powers the robot’s ongoing operations, the other can be exchanged.

Each battery provides approximately two hours of operation while walking or up to four hours when the robot stands still and performs tasks. The battery swap stations are not merely power hubs; they also meticulously monitor the health of each battery.

Should a battery show signs of degradation, a technician can be alerted to a timely replacement, further optimising the robot’s longevity and performance.

UBTech claims the Walker S2 is not a mere laboratory prototype but a robust solution engineered for real-world industrial deployment. Extensive testing has been conducted in the highly demanding environments of car factories operated by major Chinese electric vehicle manufacturers, including BYD, Nio, and Zeekr.

The trials validate the robot’s ability to operate effectively in dynamic production lines. The Walker S2 incorporates advanced vision systems, allowing it to detect battery levels and identify fully charged units, indicated by a green light on the stacked battery packs.

The robot autonomously reads the visual cues, ensuring precise selection and connection via a simple USB-style connector. Furthermore, the robot features a display face, enabling it to communicate its operational status to human workers, fostering a collaborative and transparent work environment. For safety, a prominent emergency stop button is also integrated.

China’s strategic investment in robotics is a driving force behind such innovations. Shenzhen, UBTech’s home base, is a thriving hub for robotics, boasting over 1,600 companies in the sector.

The nation’s broader push towards automation, part of its ‘Made in China 2025’ strategy, is a clear statement of global competitiveness, with China betting on AI and robotics to spearhead the next manufacturing era.

The coordinated industrial policy has led to China becoming the world’s largest market for industrial robots and a significant innovator in the field. The implications of robots like the Walker S2, built for non-stop operation, extend far beyond traditional factory floors.

Their ability to manage physical tasks continuously could redefine work in various sectors. Industries such as logistics, with vast warehouses requiring constant material handling, or airports, where baggage and cargo movement is ceaseless, benefit immensely.

Hospitals could also see these humanoids assisting with logistical duties, allowing human staff to concentrate on direct patient care. For businesses, the promise of 24/7 automation translates directly into increased output without additional human resources, ensuring operations move seamlessly day and night.

The Walker S2 exemplifies how advanced automation rapidly moves beyond research labs into practical, demanding workplaces. With its autonomous battery-swapping capability, humanoid robots are poised to work extended hours that far exceed human capacity.

The robots do not require coffee breaks or need sleep; they are designed for relentless productivity, marking a significant step towards a future where machines play an even more integral role in daily industrial and societal functions.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Z.ai unveils cheaper, advanced AI model GLM-4.5

Chinese AI startup Z.ai, formerly Zhipu, is increasing pressure on global competitors with its latest model, GLM-4.5. The company has adopted an aggressive open-source strategy to attract developers. Anyone can download and use the model without licensing fees or platform restrictions.

GLM-4.5 is designed with agentic AI, breaking tasks into smaller components for improved performance. By approaching problems step by step, the model delivers more accurate and efficient outcomes. Z.ai aims to stand out through both technical sophistication and affordability.

CEO Zhang Peng says the model runs on only eight Nvidia H20 chips, while DeepSeek’s model needs sixteen. Nvidia developed the H20 to comply with US export controls aimed at China. Reducing chip demand significantly lowers the model’s operational footprint.

Zhang said the company has enough computing power and is not seeking further hardware now. Z.ai plans to charge 11 cents per million input tokens, undercutting DeepSeek R1’s 14 cents. Output tokens will cost 28 cents per million, compared to DeepSeek’s 2.19 dollars.

Such pricing could reshape large language model deployment expectations, especially in resource-limited environments. High costs have long been a barrier to broader AI adoption. Z.ai appears to be positioning itself as a more accessible alternative.

Founded in 2019, Z.ai has raised more than 1.5 billion dollars from investors including Alibaba, Tencent, and Qiming Venture Partners. It has grown quickly from a research-focused lab to one of China’s most prominent AI contenders. A public listing in Greater China is reportedly being prepared.

OpenAI recently named Zhipu among the Chinese firms it considers strategically significant in global AI development. US authorities responded by restricting American companies from working with Z.ai. The startup has nonetheless continued to expand its model lineup and partnerships.

Chinese firms increasingly invest in open-source models, often with domestic hardware compatibility in mind. Moonshot, another Alibaba-backed company, released the Kimi K2 model. Kimi K2 has received praise for its performance in coding and mathematical tasks.

Tencent has joined the race with its HunyuanWorld-1.0 model, which is built to generate immersive 3D environments. The HunyuanWorld-1.0 can accelerate game development, virtual reality design, and simulation work. Cutting-edge features are being paired with highly efficient architectures.

Alibaba also introduced its Qwen3-Coder model to assist in code generation and debugging. Such AI tools are seeing increasing use in software engineering and education. Chinese developers are positioning themselves to compete with Western offerings such as OpenAI’s Codex and Anthropic’s Claude.

The momentum within China’s AI sector is accelerating despite geopolitical and trade restrictions. A clear shift is underway from imitation to innovation, with local startups advancing independent research. Many models are trained on China-specific datasets to optimise relevance and performance.

Z.ai’s strategy combines cost reduction, efficient chip use, and broad availability. The company can build community trust and encourage ecosystem growth by open-sourcing its tools. At the same time, pricing undercuts major rivals and could disrupt the market.

Global AI development is increasingly decentralised, with Chinese firms no longer just playing catch-up. Large-scale funding and state support are helping to close gaps in hardware and training infrastructure. Z.ai is one of several firms pushing toward greater technological autonomy.

Open-source AI development is also helping Chinese companies win favour with developers outside their borders. Many international teams are experimenting with Chinese models to diversify risk and reduce reliance on US tech. Z.ai’s GLM-4.5 is among the models gaining traction globally.

By offering a powerful, lightweight, and affordable model, Z.ai is setting a new benchmark in the industry. The combination of technical refinement and strategic pricing draws attention from investors and users. A new era of AI competition is emerging.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

Huawei challenges Nvidia with AI super server

Huawei has unveiled its most powerful AI server, the CloudMatrix 384, to challenge Nvidia’s grip on the high-performance AI infrastructure market.

The system, launched at the World AI Conference in Shanghai, uses 384 Ascend 910C chips, significantly outnumbering Nvidia’s 72 B200 GPUs in the GB200 NVL72.

Although Nvidia’s GPUs remain more powerful individually, Huawei’s design relies on stacking and high-speed chip interconnection to boost overall performance.

The company claims the CloudMatrix 384 can deliver 300 petaflops of computing power, well above Nvidia’s 180 petaflops, though it consumes nearly four times more energy.

The US recently reversed its ban on Nvidia’s H20 chip exports to China, seeking to curb Huawei’s momentum. However, ongoing reports of smuggled Nvidia GPUs raise doubts over the effectiveness of these restrictions.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

DeepSeek and others gain traction in US and EU

A recent survey has found that most US and the EU users are open to using Chinese large language models, even amid ongoing political and cybersecurity scrutiny.

According to the report, 71 percent of respondents in the US and 87 percent in the EU would consider adopting models developed in China.

The findings highlight increasing international curiosity about the capabilities of Chinese AI firms such as DeepSeek, which have recently attracted global attention.

While the technology is gaining credibility, many Western users remain cautious about data privacy and infrastructure control.

More than half of those surveyed said they would only use Chinese AI models if hosted outside China. However, this suggests that while trust in the models’ performance is growing, concerns over data governance remain a significant barrier to adoption.

The results come amid heightened global competition in the AI race, with Chinese developers rapidly advancing to challenge US-based leaders. DeepSeek and similar firms now face balancing global outreach with geopolitical limitations.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!