UNESCO initiative drives new digital platform governance frameworks in South Asia

South Asia is strengthening digital platform governance through a rights-based approach shaped by regional cooperation and international guidance.

A workshop led by UNESCO brought together policymakers, civil society and academics to align platform regulation with principles of freedom of expression and access to information.

The discussions focused on addressing governance gaps linked to misinformation, platform accountability and transparency. Participants examined national experiences and identified shared regulatory challenges, emphasising the need for coordinated regional responses instead of fragmented national measures.

An initiative that also validated regional toolkits designed for policymakers and civil society, translating global principles into practical guidance. These tools aim to support the implementation of governance frameworks that reflect local contexts while upholding international human rights standards.

The process builds on UNESCO’s Internet for Trust guidelines, reinforcing a human-centred model of digital governance. Continued collaboration across South Asia is expected to strengthen regulatory capacity and ensure that digital platforms operate with greater accountability and public trust.

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

EU funding platform drives competitiveness in strategic technologies

The European Commission has highlighted the growing impact of the Strategic Technologies for Europe Platform (STEP), which has mobilised €29 billion to strengthen innovation and competitiveness across key sectors.

An initiative that supports the development and manufacturing of critical technologies, reinforcing the Union’s strategic autonomy.

Funding has been directed toward digital and deep-tech innovation, clean technologies, biotechnology and defence, combining resources from EU programmes and Member States.

Such a coordinated approach reflects efforts to reduce strategic dependencies instead of relying on fragmented investment strategies.

The platform has also improved access to funding, with hundreds of calls and projects supported across all Member States. Tools such as the STEP Seal and the planned AI-based access systems aim to simplify processes and attract further public and private investment into high-potential projects.

Looking ahead, the initiative is shaping broader reforms, including proposals for a European Competitiveness Fund. These developments signal a continued focus on streamlining funding mechanisms while supporting innovation ecosystems and long-term economic growth across Europe.

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

New quantum threat could weaken cryptocurrency encryption systems

A new warning from Google says advances in quantum computing could weaken widely used cryptographic systems protecting cryptocurrencies and digital infrastructure. A new whitepaper suggests future quantum machines may need fewer resources than previously estimated to break elliptic curve cryptography.

The research focuses on the elliptic curve discrete logarithm problem, which underpins much of today’s blockchain security. Findings suggest quantum algorithms like Shor’s could run with fewer qubits and gates, increasing concerns about cryptographic resilience.

To address the risk, the paper recommends a transition to post-quantum cryptography, which is designed to resist quantum attacks. It also outlines short-term blockchain measures, including avoiding reuse of vulnerable wallet addresses and preparing digital asset migration strategies.

Google also introduced a responsible disclosure approach using zero-knowledge proofs to communicate vulnerabilities without exposing exploitable details.

The company says this balances transparency and security, supporting coordinated efforts across crypto and research communities to prepare for quantum threats.

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

Cloudflare adds LLM layer to client-side security detection pipeline

Cloudflare has announced two changes to its client-side security offering, making Client-Side Security Advanced available to self-serve customers and offering domain-based threat intelligence at no extra cost to all users on the free Client-Side Security bundle. The update is focused on browser-based attacks that can steal data via malicious scripts without visibly disrupting a website’s normal operation.

Cloudflare says its client-side security system assesses 3.5 billion scripts per day and monitors an average of 2,200 scripts per enterprise zone. According to the company, the product relies on browser reporting, including Content Security Policy signals, rather than scanners or application instrumentation, and requires only that traffic be proxied through Cloudflare.

A central part of the announcement is a new detection pipeline combining a Graph Neural Network (GNN) with a Large Language Model (LLM). Cloudflare says the GNN analyses the Abstract Syntax Tree of JavaScript code to identify malicious intent even when scripts are minified or obfuscated. Scripts flagged as suspicious are then passed to an open-source LLM running on Workers AI for a second-stage semantic assessment intended to reduce false positives.

Cloudflare says the GNN is tuned for high recall to identify novel and zero-day threats, but that false alarms remain a challenge at internet scale. Internal evaluation results cited by the company show that the secondary LLM layer reduced false positives in the JS Integrity threat category by nearly three times across the total analysed traffic, lowering the rate from about 0.3% to about 0.1%. On unique scripts, Cloudflare says the false-positive rate fell from about 1.39% to 0.007%.

The company also describes a recent case involving a heavily obfuscated malicious script named core.js. According to Cloudflare, the payload targeted Xiaomi OpenWrt-based home routers, altered DNS settings, and attempted to change admin passwords. Cloudflare says the script was injected through compromised browser extensions rather than by directly compromising a website, and adds that its GNN detected the malicious structure while the LLM confirmed the intent.

Cloudflare argues that the two-stage design provides structural detection via the GNN and broader semantic filtering via the LLM, enabling the company to lower the GNN decision threshold without sharply increasing alert volume. Every script flagged by the GNN is also logged to Cloudflare R2 for later auditing, which the company says helps it review cases where the LLM overrode the initial verdict.

Domain-based threat intelligence is now being made available to all Client-Side Security customers, including those not using the Advanced tier. Cloudflare says the move is partly a response to attacks seen in 2025 against smaller online shops, especially on Magento, where client-side compromises continued for days or weeks after public disclosure. By extending domain-based signals more broadly, the company says site owners can more quickly identify malicious JavaScript or suspicious connections and investigate possible compromises.

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

Funding boost for UK cities innovation sector

The UK government has pledged up to £20 million to boost the creative technology sector in the Tay Cities Region. The investment aims to support innovation in areas such as video games and virtual reality while driving economic growth.

Funding will help develop local talent and accelerate projects from early research to commercial products. The initiative focuses on strengthening collaboration between businesses, researchers and public bodies to expand opportunities across the region.

Centred around Dundee and the surrounding areas, the programme will build on an established reputation in digital industries. Universities and industry partners are expected to play a key role in delivering research, training and access to investment networks.

UK officials say the move will create jobs and open new markets, while supporting emerging applications in sectors including healthcare and education. The funding forms part of a wider national strategy to strengthen innovation and regional economies.

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

EPO strengthens industry collaboration on European patent innovation

The European Patent Office (EPO) has reinforced cooperation with industry stakeholders through discussions with the German Association of Industry IP Experts, focusing on strengthening the European patent system and supporting innovation.

A meeting that brought together representatives from major industrial actors to align priorities and explore future collaboration.

Discussions between the EPO and the stakeholders centred on enhancing technology transfer, empowering startups and fostering economic growth across Europe.

Participants emphasised the importance of inclusive engagement among patent system users instead of fragmented approaches, ensuring that innovation strategies reflect both industrial and societal needs.

The Unitary Patent system was highlighted as gaining traction, particularly among smaller entities such as SMEs, individual inventors and research organisations. Such a trend reflects broader efforts to improve accessibility and scalability within the European innovation ecosystem.

AI also featured prominently, with both sides recognising its growing role in improving efficiency and quality in patent processes.

A human-centric approach remains essential, ensuring that AI deployment supports responsible innovation while maintaining high standards in patent examination and services.

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

Technology reshapes pensions engagement

New technology is reshaping how people engage with pensions, according to Financial Conduct Authority chief executive Nikhil Rathi. Speaking in London, he highlighted the growing role of AI and digital tools in helping savers better understand their retirement finances.

Pensions dashboards are expected to give millions a clearer view of their savings, potentially driving greater engagement and behavioural change. Increased visibility may encourage actions such as consolidating pension pots or adjusting contributions.

London officials warn that stronger engagement brings risks as well as opportunities, with many consumers still lacking clear retirement plans. Policymakers aim to balance protection with flexibility, promoting informed decisions while avoiding overly restrictive systems.

Advances in AI are also enabling more personalised financial guidance, making it easier for users to explore retirement scenarios. Experts say the future of pensions will depend on integrating savings, housing and wider financial planning into a more connected system.

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

MIT uses AI to detect atomic material defects

Researchers at MIT have developed an AI model capable of identifying and quantifying atomic-scale defects in materials without damaging them. The approach aims to improve the design and performance of semiconductors, batteries, and solar cells.

The model analyses data from neutron-scattering experiments and can detect up to six different point defects simultaneously. Trained on 2,000 semiconductor materials, it analyses atomic vibrations to estimate defect types and concentrations that are hard for traditional methods to measure.

Conventional techniques such as X-ray diffraction or electron microscopy typically capture only limited aspects of material defects and often require destructive testing. The AI system uses pattern recognition to build a more complete picture, offering a non-invasive option for manufacturing quality control.

Researchers say the method could eventually be adapted to more widely used tools such as Raman spectroscopy, making industrial adoption more practical. Future work will also extend the model beyond point defects to larger structural features in materials.

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

Italy fines major bank over data protection failures

The Italian Data Protection Authority has imposed a €31.8 million fine on Intesa Sanpaolo following serious shortcomings in its handling of personal data.

The case stems from unauthorised access by an employee to thousands of customer accounts, raising concerns about internal oversight and data protection safeguards.

Investigations revealed that monitoring systems failed to detect repeated unjustified access to sensitive financial information over an extended period. The breach also involved high-risk individuals, highlighting weaknesses in risk-based controls instead of robust, targeted protection measures.

Authorities in Italy identified violations of core data protection principles, including integrity, confidentiality and accountability. Additional concerns arose from delays in notifying both regulators and affected individuals, limiting the ability to respond effectively to the incident.

The case of Intesa Sanpaolo underscores increasing regulatory scrutiny of data governance practices in the financial sector. Strengthening internal controls and ensuring timely breach reporting remain essential for maintaining trust and compliance in data-driven banking environments.

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

New Microsoft Copilot update brings deeper enterprise AI integration

Microsoft has expanded its Microsoft 365 Copilot platform with new AI capabilities designed to handle complex, multi-step workflows across workplace applications. The update integrates AI into business tools, combining multiple models to improve reasoning and execution.

A key addition is Copilot Cowork, a system designed to manage longer tasks by breaking down user requests into structured plans. It can coordinate actions across apps, manage schedules, and track progress while allowing users to intervene at each stage.

The feature is currently available through Microsoft’s Frontier programme.

Further upgrades come through Researcher, an AI tool that synthesises information from multiple sources to produce detailed reports with citations. A new Critique function uses two models, one generating outputs and the other refining them to improve accuracy and reliability.

Microsoft says the updates aim to shift Copilot from a content-generation tool to an execution-focused assistant capable of carrying out real work processes. Early adopters report better planning, scheduling, and structured deliverables in enterprise environments.

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