Interdisciplinary approaches Archives | Page 82 of 194

AIOLIA framework translates AI principles into system design

An EU-funded project, AIOLIA, is examining how Europe’s approach to trustworthy AI can be applied in practice. Principles such as transparency and accountability are embedded in the AI Act’s binding rules. Turning those principles into design choices remains difficult.

The project focuses on closing that gap by analysing how AI ethics is applied in real systems. Its work supports the implementation of AI Act requirements beyond legal text. Lessons are translated into practical training.

Project coordinator Alexei Grinbaum argues that ethical principles vary widely by context. Engineers are expected to follow them, but implications differ across systems. Bridging the gap requires concrete examples.

AIOLIA analyses ten use cases across multiple domains involving professionals and citizens. The project examines how organisations operationalise ethics under regulatory and organisational constraints. Findings highlight transferable practices without a single model.

Training is central to the initiative, particularly for EU ethics evaluators and researchers working under the AI Act framework. As AI becomes more persuasive, risks around manipulation grow. AIOLIA aims to align ethical language with daily decisions.

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

Atlas agent mode fortifies OpenAI’s ChatGPT security

ChatGPT Atlas has introduced an agent mode that allows an AI browser agent to view webpages and perform actions directly. The feature supports everyday workflows using the same context as a human user. Expanded capability also increases security exposure.

Prompt injection has emerged as a key threat to browser-based agents, targeting AI behaviour rather than software flaws. Malicious instructions embedded in content can redirect an agent from the user’s intended action. Successful attacks may trigger unauthorised actions.

To address the risk, OpenAI has deployed a security update to Atlas. The update includes an adversarially trained model and strengthened safeguards. It followed internal automated red teaming.

Automated red teaming uses reinforcement learning to train AI attackers that search for complex exploits. Simulations test how agents respond to injected prompts. Findings are used to harden models and system-level defences.

Prompt injection is expected to remain a long-term security challenge for AI agents. Continued investment in testing, training, and rapid mitigation aims to reduce real-world risk. The goal is to achieve reliable and secure AI assistance.

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

AI app Splat turns photos into colouring pages for children

Splat is a new mobile app from the team behind Retro that uses generative AI to transform personal photos into colouring pages designed for children. The app targets parents seeking creative activities, free from advertising clutter and pay-per-page websites.

Users can upload images from their camera roll or select from curated educational categories, then apply styles such as cartoon, anime or comic.

Parents guide the initial setup through simple preferences instead of a lengthy account creation process, while children can colour either on-screen or on printed pages.

Splat operates on a subscription basis, offering weekly or annual plans that limit the number of generated pages. Access to payments and settings is restricted behind parental verification, helping prevent accidental purchases by younger users.

The app reflects a broader trend in applying generative AI to child-friendly creativity tools. By focusing on ease of use and offline activities, Splat positions itself as an alternative to screen-heavy entertainment while encouraging imaginative play.

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

Digital gift cards signal TikTok Shop’s retail expansion

TikTok Shop has introduced digital gift cards as part of its wider push into e-commerce. Users can purchase cards for $10 to $500 and choose animated designs for occasions such as birthdays or weddings. Availability is currently limited to the United States.

Recipients must have a TikTok account to redeem a gift card, and the balance is added to their TikTok Wallet instantly. Users can reply with a thank-you message or send a gift card as a return gesture. The approach reinforces TikTok’s focus on social interaction alongside transactions.

The feature puts the digital shop in more direct competition with established e-commerce platforms such as Amazon and eBay, which have long offered digital gift cards. Moves into higher-end retail to broaden its ambitions. The social media powerhouse is positioning itself as a full-scale online marketplace.

Momentum has continued to build, with US sales exceeding $500 million during the Black Friday and Cyber Monday period. The results highlight rising consumer confidence in the platform’s ability to drive purchases. Engagement is increasingly translating into measurable commerce.

Further developments are planned, including video messages and an interactive unboxing experience, which are expected to be released in early 2026. Expansion continues despite uncertainty around the platform’s future in the US. Negotiations over a potential sale remain unresolved ahead of January 2026.

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

Scientists develop fast AI tool for fusion energy

Scientists have developed an AI model capable of simulating complex fusion plasma in seconds, a process that previously took hours or even days. The tool, named GyroSwin, offers a faster, more cost-effective approach to designing future fusion power plants.

Traditional five-dimensional plasma simulations, which account for spatial dimensions and particle velocities, require immense supercomputing power and long computation times.

GyroSwin uses AI to learn plasma dynamics, producing accurate simulations up to 1,000 times faster than conventional methods. Faster modelling will help optimise turbulence management, a key challenge in achieving practical fusion energy.

The AI tool preserves crucial physical information, such as fluctuation scales and sheared flows, ensuring simulations remain physically interpretable.

Researchers at UK Atomic Energy Authority, Johannes Kepler University, and Emmi AI believe GyroSwin could transform the design and operation of next-generation fusion plants, including the UK’s STEP project.

GyroSwin demonstrates how AI and supercomputing can accelerate the path towards clean, abundant fusion energy while reducing the cost and complexity of plasma modelling, paving the way for a new era of fusion innovation.

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

UN report highlights AI opportunities for small businesses

AI is increasingly helping entrepreneurs in developing countries launch, manage, and grow their businesses, according to a new UNCTAD report. Start-ups and small businesses are using AI for marketing, customer service, logistics, finance, and product design.

Large language models are enabling smaller firms to adopt AI quickly and affordably, but adoption remains uneven. Many entrepreneurs struggle to see AI’s business value, and limited skills and talent slow adoption, especially in smaller firms.

Experts emphasise that supportive ecosystems, clear governance, and skills development are essential for meaningful AI integration.

Access to affordable technology and finance also plays a crucial role. Open-source platforms, collaborations, and phased adoption- from off-the-shelf tools to in-house capabilities, help firms experiment, learn, and grow while managing risk.

UNCTAD’s report highlights the importance of policy frameworks to foster AI adoption, recommending that governments provide clear, practical rules, accessible infrastructure, and targeted training.

Entrepreneurship support centres in several countries are already helping firms identify use cases and build hands-on AI skills, bridging the gap between strategy and practical implementation.

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

Japan develops AI drones to track bears

A drone school in northeastern Japan is developing AI drones to detect bears. The system will send real-time locations to authorities via a smartphone app.

The drones have night-vision and infrared cameras and follow bears automatically once detected. Each drone flies for one hour before handing off surveillance and returning to base.

AI was trained using photos of black and brown bears from a local zoo. Officials hope it will reduce dangerous encounters and offer non-lethal ways to manage bears.

Other companies are testing similar systems in Fukushima and selling AI-equipped drones commercially. The academy aims to prevent human-bear conflicts and keep people and wildlife separate.

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

AI and cloud transform New Zealand services

Microsoft’s first New Zealand hyperscale cloud region has celebrated one year, providing local businesses and services with AI-powered tools and secure data storage. Organisations are using these capabilities to innovate, improve operations and enhance customer experiences.

Collaborations with Spark and Whakarongorau Aotearoa show AI can reduce administrative workloads and strengthen service delivery. Initiatives also support skills development, helping students, teachers and job seekers gain technology-focused credentials.

The datacentre operates sustainably, using renewable energy and waterless cooling systems to lower carbon emissions. These technologies help public services and businesses become more efficient while reducing environmental impact.

Looking ahead, Microsoft plans to expand AI adoption across New Zealand, supporting productivity, community services and responsible innovation for the next phase of digital growth.

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

NVIDIA joins US energy department Genesis mission to boost scientific discovery

The US Department of Energy has confirmed NVIDIA will join its Genesis Mission as a private-sector partner, strengthening collaboration between government, industry, and academia on advanced AI.

Launched under a recent executive order, the Genesis Mission aims to reinforce US leadership across energy, scientific research, and national security. Officials say the initiative is designed to significantly increase the productivity and impact of American science and engineering.

NVIDIA will support the Genesis Mission by providing AI and high-performance computing tools for climate modelling, manufacturing, robotics, nuclear energy, and quantum research. A memorandum of understanding sets priorities covering fission, fusion, digital twins, and autonomous laboratories.

The company is already working with the US Department of Energy on open AI models for scientific simulation. Projects also include AI-driven research in materials science, biology, and supply chains, with further collaboration planned.

The partnership builds on announcements made at NVIDIA’s GTC event in Washington. The announcements include work with Oracle on the Department’s largest scientific supercomputer at Argonne National Laboratory.

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

Android botnet Kimwolf infects nearly two million smart devices

Cybersecurity researchers have identified a large Android-based botnet capable of more than distributed denial-of-service attacks, highlighting growing risks from compromised consumer devices. The botnet, dubbed Kimwolf, is estimated to control close to two million infected systems worldwide.

The findings come from QiAnXin XLab, which said Kimwolf has infected around 1.8 million devices, mainly smart TVs, set-top boxes and tablets. Most infections were observed in Brazil, India, the US, Argentina, South Africa and the Philippines.

XLab said the infection vector remains unclear, but affected devices were linked to low-cost Android-based brands used for media streaming. Researchers noted repeated attempts to disrupt the Kimwolf, with its command-and-control infrastructure taken down several times before re-emerging.

According to the report, Kimwolf has adapted by shifting to decentralised infrastructure, including the use of Ethereum Name Service domains. Analysts also identified overlaps in code and infrastructure with AISURU, a botnet linked to record-scale DDoS attacks.

Cloudflare recently described AISURU as one of the largest robot networks observed, capable of attacks exceeding 29 terabits per second. XLab said shared infrastructure suggests both botnets are operated by the same threat group.

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