Economic

AI development by Chinese companies shifts abroad

Leading Chinese technology companies are increasingly training their latest AI models outside the country to maintain access to Nvidia’s high-performance chips, according to a report by the Financial Times. Firms such as Alibaba and ByteDance are shifting parts of their AI development to data centres in Southeast Asia, a move that comes as the United States tightens restrictions on advanced chip exports to China.

The trend reportedly accelerated after Washington imposed new limits in April on the sale of Nvidia’s H20 chips, a key component for developing sophisticated large language models. By relying on leased server space operated by non-Chinese companies abroad, tech firms are able to bypass some of the effects of US export controls while continuing to train next-generation AI systems.

One notable exception is DeepSeek, which had already stockpiled a significant number of Nvidia chips before the export restrictions took effect. The company continues to train its models domestically and is now collaborating with Chinese chipmakers led by Huawei to develop and optimise homegrown alternatives to US hardware.

Neither Alibaba, ByteDance, Nvidia, DeepSeek, nor Huawei has commented publicly on the report, and Reuters stated that it could not independently verify the claims. However, the developments underscore the increasing complexity of global AI competition and the lengths to which companies may go to maintain technological momentum amid geopolitical pressure.

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

New alliance between Samsung and SK Telecom accelerates 6G innovation

Samsung Electronics and SK Telecom have taken a significant step toward shaping next-generation connectivity after signing an agreement to develop essential 6G technologies.

Their partnership centres on AI-based radio access networks, with both companies aiming to secure an early lead as global competition intensifies.

Research teams from Samsung and SK Telecom will build and test key components, including AI-based channel estimation, distributed MIMO and AI-driven schedulers.

AI models will refine signals in real-time to improve accuracy, rather than relying on conventional estimation methods. Meanwhile, distributed MIMO will enable multiple antennas to cooperate for reliable, high-speed communication across diverse environments.

The companies believe that AI-enabled schedulers and core networks will manage data flows more efficiently as the number of devices continues to rise.

Their collaboration also extends into the AI-RAN Alliance, where a jointly proposed channel estimation technology has already been accepted as a formal work item, strengthening their shared role in shaping industry standards.

Samsung continues to promote 6G research through its Advanced Communications Research Centre, and recent demonstrations at major industry events highlight the growing momentum behind AI-RAN technology.

Both organisations expect their work to accelerate the transition toward a hyperconnected 6G future, rather than allowing competing ecosystems to dominate early development.

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

MEPs call for stronger online protection for children

The European Parliament is urging stronger EU-wide measures to protect minors online, calling for a harmonised minimum age of 16 for accessing social media, video-sharing platforms, and AI companions. Under the proposal, children aged 13 to 16 would only be allowed to join such platforms with their parents’ consent.

MEPs say the move responds to growing concerns about the impact of online environments on young people’s mental health, attention span, and exposure to manipulative design practices.

The report, adopted by a large majority of MEPs, also calls for stricter enforcement of existing EU rules and greater accountability from tech companies. Lawmakers seek accurate, privacy-preserving age verification tools, including the forthcoming EU age-verification app and the European digital identity wallet.

They also propose making senior managers personally liable in cases of serious, repeated breaches, especially when platforms fail to implement adequate protections for minors.

Beyond age limits, Parliament is calling for sweeping restrictions on harmful features that fuel digital addiction. That includes banning practices such as infinite scrolling, autoplay, reward loops, and dark patterns for minors, as well as prohibiting non-compliant websites altogether.

MEPs also want engagement-based recommendation systems and randomised gaming mechanics like loot boxes outlawed for children, alongside tighter controls on influencer marketing, targeted ads, and commercial exploitation through so-called ‘kidfluencing.’

The report highlights growing public concern, as most Europeans view protecting children online as an urgent priority amid rising rates of problematic smartphone use among teenagers. Rapporteur Christel Schaldemose said the measures mark a turning point, signalling that platforms can no longer treat children as test subjects.

‘The experiment ends here,’ she said, urging consistent enforcement of the Digital Services Act to ensure safer digital spaces for Europe’s youngest users.

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

Swiss Federal Council approves update to tax information exchange rules

The Swiss Federal Council has approved significant updates to the Ordinance on the International Automatic Exchange of Information in Tax Matters. The new rules are set to take effect across Switzerland on 1 January 2026, assuming no referendum intervenes.

The revisions expand Switzerland’s international exchange of financial account information, updating the Common Reporting Standard (CRS) and introducing the new Crypto-Asset Reporting Framework (CARF).

Crypto service providers in Switzerland will now have reporting, due diligence, and registration obligations under the AEOI Ordinance, although these provisions will not apply until at least 2027.

The updated Ordinance also extends CRS rules to Swiss associations and foundations while excluding certain accounts if specific conditions are met. Transitional measures aim to facilitate the implementation of the amended CRS and CARF by affected parties more smoothly.

Deliberations on partner states for Switzerland’s crypto data exchange have been paused by the National Council’s Economic Affairs and Taxation Committee. The CARF will become law in Switzerland in 2026, but full implementation is delayed, keeping crypto-asset rules inactive for the first year.

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

AI scribes help reduce physician paperwork and burnout

A new UCLA Health study finds that AI-powered scribe tools can reduce physicians’ documentation time and may improve work satisfaction. Conducted across 14 specialities and 72,000 patient visits, the trial tested Microsoft DAX and Nabla in real-world clinical settings.

Physicians using Nabla reduced the time spent writing each note by almost 10% compared with usual care, saving around 41 seconds per note. Both AI tools modestly improved burnout, cognitive workload, and work exhaustion, but physician oversight remains essential.

The trial highlighted several limitations, including occasional inaccuracies in AI-generated notes and a single instance of mild patient safety concern. Physicians found the tools easy to use and noted an improvement in patient engagement, with most patients being receptive.

The findings provide timely evidence as healthcare systems increasingly adopt AI scribes. Researchers emphasise that rigorous evaluation is necessary to ensure patient safety and effectiveness, and that further long-term studies across multiple institutions are recommended.

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

HP cuts thousands of jobs as AI reshapes operations

HP plans to cut between 4,000 and 6,000 jobs worldwide by fiscal 2028 as it restructures operations and expands its use of AI across product development and support services.

CEO Enrique Lores said the cuts will hit development, operations and support teams, with the plan aiming to save $1 billion over three years. The company already shed up to 2,000 roles in February under an earlier restructuring plan.

AI-enabled PCs now make up over 30% of HP’s shipments in Q4 ending 31 October, driving strong demand. However, analysts at Morgan Stanley warned that rising memory chip prices fuelled by AI data centre expansion could increase costs for consumer electronics makers.

Lores noted that HP expects the impact to be felt from the second half of fiscal 2026, though existing inventory should cover the first half.

HP projected fiscal 2026 adjusted earnings of $2.90–$3.20 per share, below expectations, with first-quarter profits also falling short of forecasts. Fourth-quarter revenue reached $14.64 billion, slightly ahead of forecasts.

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

INQUBATOR set to build a competitive quantum ecosystem over four years

Germany has launched the INQUBATOR initiative to help companies, particularly SMEs, prepare for the industrial impact of quantum computing. The four-year programme offers structured support to firms facing high entry barriers and limited access to advanced technologies.

A central feature is affordable access to quantum systems from multiple vendors, paired with workshops and hands-on training. Companies can test algorithms, assess business relevance and adapt processes without investing in costly hardware or specialist infrastructure.

The project is coordinated by the Fraunhofer Institute for Applied Solid-State Physics and is funded by the Federal Ministry of Research and Technology. It brings together several Fraunhofer institutes to guide firms from early exploration to applied solutions.

Initial pilot projects span medicine, cybersecurity, insurance and automotive sectors. These examples are intended to demonstrate measurable advantages and will be followed by an open call for further use cases across a broader range of industries.

INQUBATOR aims to reduce financial and technical obstacles while expanding quantum expertise and industrial readiness in Germany. By enabling practical experimentation, it seeks to build a competitive ecosystem of quantum-literate companies over the next four years.

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

Advantage2 pushes D-Wave’s quantum computing closer to mainstream use

Quantum computing has long been framed as a future promise, but D-Wave argues real-world use has now arrived. The company says its Advantage2 system is already running complex optimisation tasks for businesses through both cloud and on-premise deployment.

D-Wave highlights a recent physics experiment as evidence of this shift, claiming the system solved a materials-modelling problem that would take a top supercomputer nearly a million years. The result, completed in minutes, serves as a proof point of practical quantum performance.

The company says accessibility is central to its approach, emphasising that Advantage2 can be programmed in Python without specialist quantum expertise. It frames this ease of use as essential to broader adoption beyond research labs.

Industry deployments are cited across logistics, telecoms, and manufacturing. D-Wave points to scheduling gains at Pattison Food Group, network optimisation at NTT Docomo, and faster production planning at Ford Otosan as examples of measurable operational benefits.

Energy efficiency is another focus, with D-Wave stating that each of its six hardware generations draws roughly 12.5 kilowatts. The company argues that this stable power use, paired with rising performance, positions quantum systems as a lower-energy option for hard computational problems.

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

AI chatbots misidentify images they created

Growing numbers of online users are turning to AI chatbots to verify suspicious images, yet many tools are failing to detect fakes they created themselves. AFP found several cases in Asia where AI systems labelled fabricated photos as authentic, including a viral image of former Philippine lawmaker Elizaldy Co.

The failures highlight a lack of genuine visual analysis in current models. Many models are primarily trained on language patterns, resulting to inconsistent decisions even when dealing with images generated by the same generative systems.

Investigations also uncovered similar misidentifications during unrest in Pakistan-administered Kashmir, where AI models wrongly validated synthetic protest images. A Columbia University review reinforced the trend, with seven leading systems unable to verify any of the ten authentic news photos.

Specialists argue that AI may assist professional fact-checkers but cannot replace them. They emphasise that human verification remains essential as AI-generated content becomes increasingly lifelike and continues to circulate widely across social media platforms.

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

OpenAI clarifies position in sensitive lawsuit

A legal case is underway involving OpenAI and the family of a teenager who had extensive interactions with ChatGPT before his death.

OpenAI has filed a response in court that refers to its terms of use and provides additional material for review. The filing also states that more complete records were submitted under seal so the court can assess the situation in full.

The family’s complaint includes concerns about the model’s behaviour and the company’s choices, while OpenAI’s filing outlines its view of the events and the safeguards it has in place. Both sides present different interpretations of the same interactions, which the court will evaluate.

OpenAI has also released a public statement describing its general approach to sensitive cases and the ongoing development of safety features intended to guide users towards appropriate support.

The case has drawn interest because it relates to broader questions about safety measures within conversational AI systems.

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