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!
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!
Apple has been fined €98 million by Italy’s competition authority after regulators concluded that its App Tracking Transparency framework distorted competition in the app store market.
Authorities stated that the policy strengthened Apple’s dominant position while limiting how third-party developers collect advertising data.
The investigation found that developers were required to request consent multiple times for the same data processing purposes, creating friction that disproportionately affected competitors.
Regulators in Italy argued that equivalent privacy protections could have been achieved through a single consent mechanism instead of duplicated prompts.
According to the Italian authority, the rules were imposed unilaterally across the App Store ecosystem and harmed commercial partners reliant on targeted advertising. The watchdog also questioned whether the policy was proportionate from a data protection perspective under the EU law.
Apple rejected the findings and confirmed plans to appeal, stating that App Tracking Transparency prioritises user privacy over the interests of ad technology firms.
The decision follows similar penalties and warnings issued in France and Germany, reinforcing broader European scrutiny of platform governance.
Would you like to learn more aboutAI, tech and digital diplomacy? If so, ask our Diplo chatbot!
Madrid has strengthened emergency response capabilities through a new collaboration between Orange and Ericsson, integrating a dedicated slice within Orange’s 5G Standalone network.
Advanced radio access and core technologies allow emergency teams to operate on prioritised connectivity during high network demand.
Police, fire and medical services benefit from guaranteed bandwidth and low-latency communications, ensuring uninterrupted coordination during incidents.
The infrastructure by Ericsson enables dynamic switching between public 5G and emergency spectrum, supporting rapid deployment when physical networks are compromised.
Resilience remains central to the design, with autonomous power systems and redundancy maintaining operations during outages. Live video transmission from firefighters’ helmets illustrates how real time data improves risk assessment and decision making on the ground.
By combining telecom innovation with public safety needs, the initiative reinforces Madrid’s role in the emergency communications leadership of the EU and demonstrates how 5G can support critical services at scale.
Would you like to learn more aboutAI, tech and digital diplomacy? If so, ask our Diplo chatbot!
Almost 3.5 million students, staff and suppliers linked to the University of Phoenix have been affected by a data breach tied to a sophisticated cyber extortion campaign. The incident followed unauthorised access to internal systems, exposing highly sensitive personal and financial information.
Investigations indicate attackers exploited a zero-day vulnerability in Oracle E-Business Suite, a widely used enterprise financial application. The breach surfaced publicly after the Clop ransomware group listed the university on its leak site, prompting internal reviews and regulatory disclosures.
Compromised data includes names, contact details, dates of birth, social security numbers and banking information. University officials have confirmed that affected individuals are being notified, while filings with US regulators outline the scale and nature of the incident.
The attack forms part of a broader wave of intrusions targeting American universities and organisations using Oracle platforms. As authorities offer rewards for intelligence on Clop’s operations, the breach highlights growing risks facing educational institutions operating complex digital infrastructures.
Would you like to learn more aboutAI, tech and digital diplomacy? If so, ask our Diplo chatbot!
France’s national postal service, La Poste, suffered a cyber incident days before Christmas that disrupted websites, mobile applications and parts of its delivery network.
The organisation confirmed a distributed denial of service attack temporarily knocked key digital systems offline, slowing parcel distribution during the busiest period of the year.
A disruption that also affected La Banque Postale, with customers reporting limited access to online banking and mobile services. Card payments in stores, ATM withdrawals, and authenticated online payments continued to function, easing concerns over wider financial instability.
La Poste stated there was no evidence of customer data exposure, although several post offices in France operated at reduced capacity. Staff were deployed to restore services while maintaining in-person banking and postal transactions where possible.
The incident added to growing anxiety over digital resilience in critical public services, particularly following a separate data breach disclosed at France’s Interior Ministry last week. Authorities have yet to identify those responsible for the attack on La Poste.
Would you like to learn more aboutAI, tech and digital diplomacy? If so, ask our Diplo chatbot!
The US tech giant NVIDIA has largely remained shut out of China’s market for advanced AI chips, as US export controls have restricted sales due to national security concerns.
High-performance processors such as the H100 and H200 were barred, forcing NVIDIA to develop downgraded alternatives tailored for Chinese customers instead of flagship products.
A shift in policy emerged after President Donald Trump announced that H200 chip sales to China could proceed following a licensing review and a proposed 25% fee. The decision reopened a limited pathway for exporting advanced US AI hardware, subject to regulatory approval in both Washington and Beijing.
If authorised, the H200 shipments would represent the most powerful US-made AI chips permitted in China since restrictions were introduced. The move could help NVIDIA monetise existing H200 inventory while easing pressure on its China business as it transitions towards newer Blackwell chips.
Strategically, the decision may slow China’s push for AI chip self-sufficiency, as domestic alternatives still lag behind NVIDIA’s technology.
At the same time, the policy highlights a transactional approach to export controls, raising uncertainty over long-term US efforts to contain China’s technological rise.
Would you like to learn more aboutAI, tech and digital diplomacy? If so, ask our Diplo chatbot!
Google Cloud’s 2026 AI Agent Trends Report shows AI agents are moving from experimental tools to central business systems. Employees are shifting from routine execution to oversight and strategic decision-making.
The report highlights agents managing end-to-end workflows across teams, thereby improving efficiency and streamlining complex processes. Personalised customer service is becoming faster and more accurate thanks to these systems.
Security operations are seeing benefits as AI agents handle alerts, investigations and fraud detection more effectively. Human analysts can now focus on higher-value tasks while routine work is automated.
Companies are investing in continuous training to build an AI-ready workforce. The report emphasises that people, not just technology, will determine the success of AI adoption.
Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!
Ghana has formally legalised Bitcoin and cryptocurrency trading after parliament approved the Virtual Asset Service Providers Bill, 2025, closing a long-standing regulatory gap in the country’s digital asset market.
The legislation establishes a licensing and supervisory regime for crypto businesses under the Bank of Ghana. The central bank will oversee the sector, prioritising consumer protection and financial stability, while unlicensed operators may face sanctions or closure.
Under the new framework, individuals can trade crypto legally, while companies must meet reporting and compliance requirements. Officials say the law responds to fraud and money laundering risks while acknowledging the scale of crypto adoption nationwide.
Around 3 million Ghanaians have used cryptocurrency, with transactions totalling roughly $3 billion by June 2024. Licensing rules will be introduced gradually in 2026, as Ghana aligns with a broader African shift toward formal crypto regulation.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
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!
