The European Commission has updated the ERA Living Guidelines on the responsible use of generative AI in research, reflecting the growing use of AI tools across scientific work. The revised guidance aims to support researchers, research organisations and funding bodies in adopting generative AI while maintaining core principles of research integrity.
The guidelines emphasise reliability, honesty, respect and accountability, including transparency over AI use, protection of privacy and confidential information, and responsibility for research outputs. They also stress that researchers remain ultimately responsible for scientific output and should verify AI-generated results.
New recommendations address risks linked to the use of generative AI by third parties, including in meetings, note-taking, summaries and document overviews, where confidential information, data protection or intellectual property rights may be affected. The guidelines encourage researchers and organisations to inform third parties about the use of such tools and related risks.
A specific addition concerns the risk of ‘hidden prompts’, where instructions may be secretly embedded in documents or inputs to influence generative AI tools. The guidelines call on research funding organisations to remain aware of such risks, set rules prohibiting manipulation where relevant, and introduce appropriate safeguards in IT systems used to process information.
Developed through the European Research Area Forum, the guidelines are intended as a non-binding supporting tool for the research community. The Commission says they will be updated regularly and that users can continue to provide feedback as generative AI and the surrounding policy landscape evolve.
Why does it matter?
Generative AI is becoming part of everyday research workflows, from drafting and summarising to proposal preparation and document analysis. The updated guidelines show that research integrity risks now extend beyond individual misuse to organisational processes, third-party tools and hidden technical behaviours that may affect scientific judgement. Shared guidance across the European Research Area can help institutions adopt AI without weakening transparency, accountability or trust in research.
Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our chatbot!
A new study co-authored by economists from Massachusetts Institute of Technology and Yale University finds that automation in the United States has often been driven less by productivity gains and more by firms’ efforts to reduce labour costs.
Rather than replacing workers to maximise efficiency, companies have frequently targeted employees earning a ‘wage premium’, effectively lowering higher-than-average salaries within comparable roles.
The research suggests this pattern has contributed significantly to widening income inequality while delivering only limited productivity improvements.
The analysis, which examines data spanning multiple decades and industries, indicates that automation has disproportionately affected higher-earning workers within affected groups. It also estimates that inefficient automation deployment may have offset a large share of potential productivity gains over time.
Researchers argue that the findings highlight a structural tension in how automation is applied, where short-term cost reduction can take priority over long-term economic efficiency, shaping both wage distribution and overall growth dynamics in the US economy since 1980.
Why does it matter?
The findings challenge the assumption that automation primarily improves efficiency and productivity, showing instead that firms can strategically use it to reshape wage structures and concentrate economic gains.
From a broader perspective, this helps explain why technological progress has not translated evenly into higher productivity or shared prosperity, while also highlighting how corporate incentives can steer innovation in ways that deepen inequality across labour markets.
Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!
The New South Wales Civil and Administrative Tribunal has issued guidance on the acceptable use of generative AI in tribunal proceedings as part of Privacy Awareness Week NSW 2026, which this year focuses on personal information risks in the age of AI.
According to NCAT, generative AI tools may be used to assist with administrative and organisational tasks such as summarising material, organising information, or preparing chronologies. At the same time, the tribunal warns that such tools can create privacy risks if users enter personal, sensitive, or confidential information.
The guidance is set out in NCAT Procedural Direction 7 on the use of generative AI, together with an accompanying fact sheet. NCAT says the aim is to clarify when generative AI may be used in tribunal-related work while reinforcing obligations to protect personal and confidential information.
The tribunal also draws a clear line around evidentiary material. Generative AI must not be used to generate or alter evidence in tribunal proceedings, including statements, affidavits, statutory declarations, character references, or other evidentiary documents.
NCAT further states that generative AI must not be used to generate content for an expert report unless the tribunal has given permission. It is encouraging parties and their representatives to review the guidance before using such tools in proceedings.
Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!
The International Labour Organization has warned that governments must place lifelong learning at the centre of economic and social policy as AI, digitalisation and demographic shifts continue transforming labour markets worldwide. The organisation said stronger and more inclusive learning systems are necessary to prevent widening inequality between workers, industries and countries.
According to the ILO’s new report, titled ‘Lifelong learning and skills for the future’, only 16% of people aged between 15 and 64 participated in structured training during the previous year. Access remains significantly higher among full-time employees in formal companies, where employer-supported training reaches 51%.
The ILO report warns that workers in informal jobs and smaller enterprises continue relying mainly on learning through experience instead of structured education programmes. Furthermore, the study found that employers increasingly seek combinations of digital, socio-emotional, communication and problem-solving skills rather than narrow technical expertise alone.
While demand for AI-related capabilities is expected to increase, the report noted that most workers currently use ready-made AI tools that require broader digital literacy, critical thinking and collaborative abilities instead of specialist engineering knowledge.
The ILO also highlighted the growing importance of green and care economy skills. It estimates that 32% of workers globally already perform environmentally relevant tasks, while demand for long-term care workers could almost double by 2050.
The organisation called for greater public investment, stronger institutional coordination and inclusive lifelong learning strategies capable of supporting workers throughout rapidly changing technological and economic transitions.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
DeepSeek has again placed itself at the centre of the global AI race. After drawing worldwide attention with its R1 reasoning model in early 2025, the Chinese company has recently released DeepSeek V4, a new model designed to compete not only on performance, but also on price, openness and efficiency.
The hype around DeepSeek V4 is not based on a single feature. The model comes with a 1 million-token context window, open weights, two versions for different use cases and a strong focus on agentic workflows such as coding, research, document analysis and long-running tasks. In a market still dominated by expensive closed models, DeepSeek is trying to prove that powerful AI does not need to remain locked behind trademarked systems.
A model built for long memory
The most immediate difference between DeepSeek V4 and other models is context length. Both DeepSeek-V4-Pro and DeepSeek-V4-Flash support a 1-million-token context window, meaning they can process inputs far longer than those of older generations of mainstream models. According to DeepSeek’s official release, one million tokens is now the default across all official DeepSeek services.
For ordinary users, that may sound technical. In practice, it matters because a longer context allows models to work with large documents, long conversations, full codebases, legal materials, research archives or complex project histories without losing track as quickly.
That is why DeepSeek V4 is not just another chatbot release. It is aimed at the next stage of AI use, where models are expected to act less like question-answering tools and more like assistants that can follow long processes over time.
Two models for two different needs
DeepSeek V4 comes in two main versions. DeepSeek-V4-Pro is a larger and more capable model, with 1.6 trillion total parameters and 49 billion active parameters. DeepSeek-V4-Flash is a smaller model, with 284 billion total parameters and 13 billion active parameters, designed for faster and more cost-effective workloads.
That distinction is important. Not every user needs the strongest model for every task. A company summarising documents, routing queries or running basic support may choose Flash. A developer working on complex coding tasks, long-context agents or advanced reasoning may prefer Pro.
DeepSeek’s release reflects a broader trend in AI. The best model is no longer always the biggest one. Cost, speed, context size and deployment flexibility are now as important as raw benchmark performance.
Why the price matters
One reason DeepSeek attracts so much attention is its aggressive pricing. DeepSeek’s API page lists V4-Flash at USD 0.14 per 1 million input tokens on a cache miss and USD 0.28 per 1 million output tokens. V4-Pro is listed at USD 1.74 per 1 million input tokens and USD 3.48 per 1 million output tokens before the temporary 75% discount.
For developers and companies, that changes the calculation. High-performing AI models are useful only if they can be deployed at scale. If every long document, coding session or agentic workflow becomes too expensive, adoption slows down.
DeepSeek’s challenge to the market is therefore not only technical. It is economic. The company is pushing the idea that frontier-level AI should be cheaper to run, easier to access and less dependent on closed ecosystems.
The architecture behind the hype
DeepSeek V4 uses a mixture-of-experts approach, meaning only part of the model is active during each response. That helps explain why the model can be very large on paper, yet still more efficient to run than a dense model of similar overall size.
The more interesting part is how DeepSeek handles long context. NVIDIA’s technical overview explains that DeepSeek V4 uses hybrid attention, combining compression and selective attention techniques to reduce the cost of processing very long prompts. NVIDIA says these changes are designed to cut per-token inference FLOPs by 73% and reduce KV cache memory burden by 90% compared with DeepSeek-V3.2.
For a non-technical audience, the point is simple. DeepSeek V4 is trying to solve one of the biggest problems in modern AI: how to make models remember and process much more information without becoming too slow or too expensive.
That is where much of the hype comes from. The model is not merely larger. It is designed around the economics of long-context AI.
Why NVIDIA is still in the picture
NVIDIA’s role in the DeepSeek V4 story is especially interesting. DeepSeek is often discussed as part of China’s effort to build a more independent AI ecosystem, but NVIDIA has also been quick to move forward to support developers who want to build with the model.
In its technical blog, NVIDIA describes DeepSeek V4 as a model family designed for efficient inference of million-token contexts. The company says DeepSeek-V4-Pro and V4-Flash are available through NVIDIA GPU-accelerated endpoints, while developers can also use NVIDIA Blackwell, NIM containers, SGLang and vLLM deployment options.
NVIDIA also reports that early tests of DeepSeek-V4-Pro on the GB200 NVL72 platform showed more than 150 tokens per second per user. That matters because long-context models place heavy memory pressure, as well as on compute and networking infrastructure. The model may be efficient by design, but serving it at scale still requires serious hardware.
So, DeepSeek V4 does not remove NVIDIA from the story – it complicates it. The model is part of a broader push towards more efficient AI, but the infrastructure race remains central.
The chip question behind the model
DeepSeek V4 also arrives at a time when AI infrastructure is becoming just as important as model performance. MIT Technology Review frames the release partly through that lens, noting that DeepSeek’s new model reflects China’s broader attempt to reduce reliance on foreign AI hardware and build a more self-sufficient technology stack.
That detail matters because the AI race is no longer only about who builds the most capable model. It is also about who controls the chips, software frameworks and data centres needed to run it.
Replacing NVIDIA, however, remains difficult. Its advantage lies not just in its chips, but also in the software ecosystem developers have built around its platforms over many years. Moving to alternative hardware means adapting code, rebuilding tools and proving that the new systems are stable enough for serious use.
DeepSeek V4, however, sits between two realities. It points towards China’s ambition to build a more independent AI stack, while NVIDIA’s rapid support for the model shows that frontier AI still depends heavily on established infrastructure.
Open weights as a strategic move
DeepSeek V4 is also important because the model weights are available through Hugging Face under the MIT License. That gives developers more freedom to inspect, adapt and deploy the model than they would have with a fully closed commercial system.
Open-weight models are becoming a major pressure point in the AI race. Closed models may still lead in some areas, especially in polished consumer products, enterprise support and safety layers. However, open models offer something different: flexibility.
For universities, start-ups, smaller companies and developers outside the largest AI ecosystems, that flexibility matters. It means advanced AI can be tested, modified and integrated without relying entirely on a handful of dominant providers.
Benchmarks need caution
DeepSeek presents V4-Pro as highly competitive across reasoning, coding, long-context and agentic benchmarks. Hugging Face lists results including 80.6 on SWE-bench Verified, 90.1 on GPQA Diamond and 87.5 on MMLU-Pro for DeepSeek-V4-Pro.
Those numbers are impressive, but they should not be treated as the full story. Benchmarks are useful, but they rarely capture every real-world use case. A model can score well on coding tests and still struggle with reliability, factual accuracy, safety or complex multi-step workflows in production.
That caution is important. The AI industry often turns benchmarks into headlines, while real performance depends on deployment, prompting, safety controls and the specific task at hand.
More than just another model release
DeepSeek V4 matters because it combines several trends into one release: long context, lower prices, open weights, agentic workflows and geopolitical competition. It also shows that the AI race is no longer fought only in labs, benchmarks and data centres. Visibility now matters too. Tools such as Diplo’s Digital Footprints show how digital presence shapes the way technology actors and media narratives are discovered, ranked and understood. At this stage, the competition is not only about who has the smartest model. It is also about who can make intelligence cheaper, more available and easier to deploy.
That does not mean DeepSeek has solved every problem. Questions remain around independent benchmarking, safety, data governance, infrastructure and the broader political context of Chinese AI development. Still, the release does show where the market is heading.
The next phase of AI may not be defined solely by the most powerful model. It may be defined by the model that is powerful enough, affordable enough and open enough to change how people build products, services and tools with AI.
Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!
The Department said the platform brings together resources to support AI literacy and structured AI-focused training pathways across sectors.
The portal is organised around three main areas: AI skills integration in apprenticeships, industry-specific training modules, and pathways for embedding AI into both new and existing programmes.
The Department said training content spans sectors including healthcare, finance, education, construction, advanced manufacturing and technology.
Alongside the portal, the Department has introduced an AI Literacy Framework to guide employers, educators and training providers. The Department said the AI Literacy Framework outlines core competencies, including understanding AI capabilities and limits, using tools in daily tasks, and assessing output accuracy.
A separate initiative, the Make America AI-Ready programme, delivers a free text-message-based AI course aimed at workers without reliable internet access.
Officials said organisations can join existing apprenticeships, create new AI-focused schemes, or update current programmes to include AI skills. The project aligns with wider federal strategies to accelerate AI education and workforce readiness across the United States.
Why does it matter?
The initiative signals a structural shift in how governments are preparing the workforce for AI integration, embedding practical skills into formal apprenticeship systems rather than treating them as optional add-ons.
It also broadens access to AI literacy by targeting both high-growth industries and digitally excluded workers, helping reduce future gaps in productivity and employability.
Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!
Digital Dubai has launched the AI Workforce Transformation Programme to train 50,000 government employees in AI skills. The initiative is being delivered with the Dubai Government Human Resources Department and the Dubai Centre for Artificial Intelligence.
The programme aims to equip staff with practical knowledge to apply AI in public services and internal processes. It includes tailored training tracks based on job roles, from leadership to general employees.
Officials say the initiative will improve productivity, support innovation and enable more efficient service delivery. It also forms part of wider efforts to strengthen AI adoption across government operations.
The programme is designed to build long-term institutional capabilities and support a technology-driven government model. The initiative was launched by Digital Dubai in Dubai.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
GovTech Singapore has argued that stronger AI governance in workplaces is essential for trust, compliance, risk management, and responsible innovation as AI adoption expands across business operations.
The agency leading Singapore’s Smart Nation and digital government efforts defines AI governance as a framework of policies, processes, and responsibilities guiding the ethical, transparent, and accountable development and deployment of AI systems within an organisation. The framework is linked to oversight across the AI lifecycle, from design through to ongoing monitoring.
Key elements identified by GovTech Singapore include transparency and explainability, fairness and bias mitigation, accountability and human oversight, and data privacy and security. Responsible AI is also linked to Singapore’s wider Smart Nation agenda, which the agency describes as a national priority.
The guidance recommends that organisations establish clear internal policies on AI use, build AI literacy across teams, carry out regular audits and assessments, and prioritise secure development practices. It also points to Singapore’s Model AI Governance Framework for Generative AI, developed by the AI Verify Foundation and the Infocomm Media Development Authority, as a reference point for businesses adapting governance frameworks to their own needs.
As part of its effort to support responsible AI use in the public sector, GovTech Singapore also highlights its AI Guardian suite. The suite includes Litmus, a testing platform using adversarial prompts to identify risks and vulnerabilities, and Sentinel, a guardrails service designed to detect and mitigate unsafe or irrelevant content before it affects AI models or users.
Overall, GovTech Singapore presents AI governance not only as a compliance issue, but as part of building a trusted digital environment in which AI can be deployed safely and effectively.
Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!
Kazakhstan has introduced an AI-powered assistant designed to simplify the process of starting a business, according to Zhaslan Madiyev. Developed in cooperation with the Ministry of Finance, the platform aims to provide data-driven guidance to early-stage entrepreneurs.
Built around a digital mapping system, the assistant evaluates factors such as nearby businesses, customer flow, and competition. Its recommendations aim to help users choose more viable locations and avoid oversaturated sectors, thereby reducing the risk of duplicating businesses in the same area.
Officials say the tool could reduce startup operating costs by up to half while improving long-term business sustainability. Alongside it, a second AI assistant already provides continuous guidance on tax reporting and regulatory compliance, translating complex requirements into clearer, more practical steps for users. According to Kazakhstani reporting, the tax assistant has already processed more than 5,000 requests.
The development forms part of Kazakhstan’s wider digital transformation agenda, which aims to modernise public services and strengthen the country’s digital economy through practical AI deployment. The government says more than 50 AI-powered services are now being developed to support citizens and businesses.
Why does it matter?
Kazakhstan’s AI assistant points to a shift from basic digital services towards more active, real-time decision support for entrepreneurs. Data-driven recommendations can help reduce startup risks, limit market oversaturation, and support more efficient resource allocation across local economies.
Simplified tax and compliance guidance also targets one of the main barriers facing early-stage businesses: administrative complexity. Placed within Kazakhstan’s broader AI-first digital strategy, the initiative signals a wider move towards a more competitive and operationally AI-driven digital economy.
Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!
Malaysia is strengthening its national AI strategy through an expanded partnership with Microsoft, launching the Microsoft Elevate initiative to accelerate AI readiness across society.
An initiative that targets educators, public sector institutions, small businesses and wider communities, aiming to embed practical AI capabilities into everyday economic and social activity.
Early deployment has already reached tens of thousands of learners, reflecting a shift from pilot programmes to large-scale national implementation.
Government and industry leaders in Malaysia emphasise that long-term competitiveness depends not only on technological investment but on widespread adoption and understanding of AI tools.
The programme therefore prioritises workforce activation, institutional capacity and sustainable integration across sectors.
Malaysia’s approach reflects a broader global trend where public–private partnerships are increasingly central to AI development, focusing on inclusive access, responsible use and real-world application rather than purely technological advancement.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
