Capacity development Archives | Digital Watch Observatory

Bulgaria becomes first country to deploy a national AI model across a tax authority

Bulgaria’s National Revenue Agency (NRA) has begun rolling out an AI system developed by INSAIT, the Institute for Computer Science, Artificial Intelligence and Technology at Sofia University, across all of its organisational structures, making it the first large-scale public administrative body in the country to deploy the BgGPT national language model.

Following a successful pilot phase, the system is now in expanded use across the NRA’s central office and seven territorial directorates.

The AI system enables staff to conduct general and specialised searches related to tax and social security legislation, generating instant responses to improve service quality for citizens and businesses.

Crucially, it runs exclusively on open-weight models and operates on proprietary hardware, an approach specifically designed to prevent data leakage and protect privacy, two of the central concerns when integrating AI into government institutions.

The next phase of the project will see the system adapted for specialised use cases and integrated into internal processes alongside national integrator ‘Information Services’, with the goal of reaching daily use by more than 7,000 NRA employees.

INSAIT describes the initiative as a concrete contribution to European AI sovereignty, with Bulgaria combining nationally developed language models and locally controlled hardware to reduce dependence on commercial AI providers.

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

Mastercard expands AI strategy with new payments model

Mastercard has introduced a generative AI foundation model trained on billions of anonymised transactions. The model is designed as a backend system to power insights across payments and commerce services.

The company plans to extend AI use beyond fraud detection into cybersecurity, loyalty programmes and small-business tools. The model is being developed with support from Nvidia and Databricks technologies.

Earlier AI tools focused on fraud detection, significantly improving accuracy and reducing false positives. The new model marks a shift towards a broader infrastructure approach across multiple products.

This move aligns with Mastercard’s growing reliance on value-added services, which generated over $13 billion in revenue. These services include security, analytics and digital payment solutions beyond the core network.

Competitors such as Visa and PayPal are also expanding AI-driven commerce platforms. The race is intensifying as firms build integrated systems for payments, automation and intelligent services.

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

Agentic Ready programme by Visa launched to prepare for AI-driven payments

Visa has launched Agentic Ready, a global programme preparing the payments ecosystem for AI agents to initiate transactions for consumers. The programme builds on Visa Intelligent Commerce, the company’s framework for secure, AI-driven payment experiences.

The first phase, launching in Europe, including the UK, focuses on issuer readiness. Participating banks and financial institutions can test and validate agent-initiated transactions in controlled production environments, ensuring they remain secure, reliable, and scalable.

Visa’s trust layer integrates tokenisation, identity verification, risk controls, and biometric authentication to maintain consumer consent and protection throughout transactions.

Controlled testing with selected merchants allows issuers to gain practical experience of agentic commerce in real-world settings. Early participants, including Barclays, HSBC UK, Revolut, and Banco Santander, help Visa test and refine safe AI-driven payments across channels.

The programme advances Visa’s vision of AI-driven commerce, enabling flexible payments while keeping consumers in control. Expansion beyond Europe is planned, leveraging lessons from the initial rollout to accelerate agentic commerce globally.

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

Horizon Worlds remains active as Meta reconsiders VR plans

Meta has reversed its earlier decision to discontinue virtual reality support for Horizon Worlds, allowing the platform to remain available on VR headsets despite previous plans to prioritise mobile and web access.

The decision follows an internal reassessment of user engagement trends, which indicate limited adoption of VR-based social platforms.

While Horizon Worlds was once positioned as central to the company’s metaverse ambitions, demand has remained relatively low, raising questions about the long-term viability of immersive social environments.

Financial pressures also continue to shape strategy.

Meta’s Reality Labs division has recorded substantial losses since 2021, reflecting high investment in virtual and augmented reality technologies without corresponding commercial returns.

Industry data further suggests declining headset sales, reinforcing uncertainty around VR as a mainstream consumer platform.

In contrast, mobile usage of Horizon Worlds is growing faster. Increasing downloads point to broader accessibility and improved product-market alignment, though revenue generation remains limited.

As a result, Meta is prioritising mobile development instead of fully abandoning VR, maintaining a dual approach while seeking more sustainable engagement models.

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

UNESCO promotes safe AI use and gender equality in Caribbean workshop

A regional workshop in Kingston has been organised by UNESCO to explore the relationship between AI, gender equality and online safety, reflecting wider efforts to support inclusive digital governance across the Caribbean.

Discussions examined the impact of technology-facilitated gender-based violence, including harassment, impersonation and image-based abuse, which continue to affect women and girls disproportionately.

Generative AI was presented as both an opportunity and a risk, with concerns linked to bias, deepfakes, misinformation and non-consensual content.

More than 50 participants from government, civil society and youth organisations engaged in practical sessions aimed at strengthening awareness and digital skills. A participatory approach encouraged peer learning and critical thinking, aligning with UNESCO’s ethical AI principles.

‘Technology reflects the hands that build it and the society that feeds it data. If we are not careful, AI will not just mirror our existing inequalities; it will magnify them.‘

The Honourable Olivia Grange, Minister of Culture, Gender, Entertainment and Sport of Jamaica.

‘The pursuit of equality must extend into every space where women live, work, and where they connect and express themselves – including the digital world,‘

For Eric Falt, Regional Director and Representative of UNESCO.

The initiative forms part of broader efforts to ensure that digital transformation supports inclusion rather than reinforcing existing disparities, while equipping stakeholders with tools for safe and responsible AI use.

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

Bitcoin moves closer to quantum resistance with BIP-360

BTQ Technologies has deployed Bitcoin Improvement Proposal BIP-360 on its Bitcoin Quantum Testnet v0.3.0, marking the first live test of the proposal. The upgrade introduces a quantum-resistant transaction model, Pay-to-Merkle-Root, designed to strengthen Bitcoin’s long-term security.

BIP-360 focuses on mitigating a vulnerability linked to Taproot’s key-path spending mechanism, which can expose public keys on-chain. Such exposure may become a risk if future quantum computers are capable of exploiting cryptographic weaknesses using advanced algorithms.

The testnet adds new consensus rules, post-quantum signatures, and full transaction lifecycle testing. Faster one-minute block times and adjusted fee structures have been introduced to accommodate larger and more complex signatures.

Growing global attention on quantum threats adds urgency to the development. US, EU, and Canadian authorities are setting timelines for post-quantum cryptography to protect future system security.

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

EU scrutiny intensifies over Broadcom VMware licensing dispute

Broadcom is facing increased regulatory pressure in the EU following a formal antitrust complaint concerning changes to VMware licensing practices.

The complaint highlights growing tensions between large technology providers and European cloud infrastructure firms.

The filing, submitted by Cloud Infrastructure Services Providers in Europe, raises concerns that revised licensing models could significantly alter market dynamics.

European providers argue that the changes may limit flexibility, increase costs, and affect their ability to compete effectively in the cloud services sector.

At the centre of the dispute lies the broader issue of market concentration and control over critical digital infrastructure.

Industry stakeholders suggest that restrictive licensing conditions could reshape access to essential virtualisation technologies, which underpin a wide range of cloud and enterprise services across the EU.

Regulatory attention is expected to focus on whether such practices align with the EU competition rules, particularly regarding fair access and market neutrality.

The case emerges at a time when European policymakers are intensifying oversight of dominant technology firms and seeking to strengthen digital sovereignty across strategic sectors.

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

MIT develops method to detect overconfident AI

Researchers at MIT have introduced a new method to assess the reliability of large language models more accurately. Many LLMs can produce confident yet incorrect responses, posing risks in high-stakes applications such as healthcare or finance.

The team combined self-consistency checks with an ensemble approach, comparing a model’s outputs to similar LLMs. This total uncertainty (TU) metric more accurately identifies overconfident predictions and can flag hallucinations that simpler methods may miss.

Experiments on ten common tasks- including question-answering, translation, summarisation, and math reasoning- showed that TU outperformed individual uncertainty measures.

The ensemble approach relies on models from different developers to ensure diversity and credibility, offering a practical and energy-efficient way to gauge AI confidence.

Researchers suggest TU could also help reinforce correct answers during training, improving overall model performance. Future developments aim to enhance the metric’s accuracy for open-ended tasks and explore additional forms of uncertainty.

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

AI safety push sees Anthropic and OpenAI recruit explosives specialists

Anthropic and OpenAI are recruiting chemical and explosives experts to strengthen safeguards for their AI systems, reflecting growing concern about the potential misuse of advanced models.

Anthropic is seeking a policy specialist to design and monitor guardrails governing how its systems respond to prompts involving chemical weapons and explosives. The role includes assessing high-risk scenarios and responding to potential escalation signals in real time.

OpenAI is expanding its Preparedness team, hiring researchers and a threat modeller to identify and forecast risks linked to frontier AI systems. The positions focus on evaluating catastrophic risks and aligning technical, policy, and governance responses.

The recruitment drive comes amid heightened scrutiny of AI safety and national security implications. Anthropic is currently challenging a US government designation that labels it a supply-chain risk, while tensions have emerged over restrictions on the military use of AI systems.

At the same time, OpenAI has secured agreements to deploy its technology in classified environments under defined constraints. The parallel developments highlight how AI firms are balancing commercial expansion with increasing pressure to implement robust safety controls.

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

Data centres drive LG’s integrated AI infrastructure push

AI infrastructure is becoming a central battleground for growth, with LG Group accelerating its push into AI data centres and energy storage systems under its ‘One LG’ strategy.

The initiative brings together key affiliates to deliver integrated solutions for AI data centres. LG Electronics provides cooling systems, LG Energy Solution handles power infrastructure, including ESS and UPS, while LG Uplus and LG CNS oversee design, construction, and operations.

The strategy comes as global demand for AI data centres surges, driven by energy-intensive workloads and rising electricity constraints. Expanding storage capacity has become critical, with the US expected to add over 24 gigawatts of energy storage capacity in 2026 alone.

LG Electronics is focusing on advanced cooling technologies, including large air-cooled chillers and liquid-cooling systems, to manage the intense heat generated by GPU-intensive AI workloads. The company has also expanded into immersion cooling through partnerships, aiming to achieve efficiency gains in next-generation facilities.

Meanwhile, LG Energy Solution is strengthening its role in power infrastructure, scaling ESS production across North America, and securing major contracts. Through integrated battery and software solutions, the company is positioning itself to meet growing demand for stable, high-capacity energy systems supporting AI operations.

On the networking side, LG Uplus is developing low-latency infrastructure and AI-driven data centre management systems to optimise performance and energy use in real time. Together, these efforts highlight LG’s ambition to become a full-stack provider in the rapidly expanding AI data centre ecosystem.

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