Human rights principles

Chinese court limits liability for AI hallucinations

A court in China has ruled that AI developers are not automatically liable for hallucinations produced by their systems. The decision was issued by the Hangzhou Internet Court in eastern China and sets an early legal precedent.

Judges found that AI-generated content should be treated as a service rather than a product in such cases. In China, users must therefore prove developer fault and show concrete harm caused by the erroneous output.

The case involved a user in China who relied on AI-generated information about a university campus that did not exist. The court ruled no damages were owed, citing a lack of demonstrable harm and no authorisation for the AI to make binding promises.

The Hangzhou Internet Court warned that strict liability could hinder innovation in China’s AI sector. Legal experts say the ruling clarifies expectations for developers while reinforcing the need for user warnings about AI limitations.

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

Moltbook AI vulnerability exposes user data and API keys

A critical security flaw has emerged in Moltbook, a new AI agent social network launched by Octane AI.

The vulnerability allowed unauthenticated access to user profiles, exposing email addresses, login tokens, and API keys for registered agents. The platform’s rapid growth, claimed to have 1.5 million users, was largely artificial, as a single agent reportedly created hundreds of thousands of fake accounts.

Moltbook enables AI agents to post, comment, and form sub-communities, fostering interactions that range from AI debates to token-related activities.

Analysts warned that prompt injections and unregulated agent interactions could lead to credential theft or destructive actions, including data exfiltration or account hijacking. Experts described the platform as both a milestone in scale and a serious security concern.

Developers have not confirmed any patches, leaving users and enterprises exposed. Security specialists advised revoking API keys, sandboxing AI agents, and auditing potential exposures.

The lack of safeguards on the platform highlights the risks of unchecked AI agent networks, particularly for organisations that may rely on them without proper oversight.

An incident that underscores the growing need for stronger governance in AI-powered social networks. Experts stress that without enforced security protocols, such platforms could be exploited at scale, affecting both individual users and corporate systems.

The Moltbook case serves as a warning about prioritising hype over security in emerging AI applications.

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

Grok returns to Indonesia as X agrees to tightened oversight

Indonesia has restored access to Grok after receiving guarantees from X that stronger safeguards will be introduced to prevent further misuse of the AI tool.

Authorities suspended the service last month following the spread of sexualised images on the platform, making Indonesia the first country to block the system.

Officials from the Ministry of Communications and Digital Affairs said that access had been reinstated on a conditional basis after X submitted a written commitment outlining concrete measures to strengthen compliance with national law.

The ministry emphasised that the document serves as a starting point for evaluation instead of signalling the end of supervision.

However, the government warned that restrictions could return if Grok fails to meet local standards or if new violations emerge. Indonesian regulators stressed that monitoring would remain continuous, and access could be withdrawn immediately should inconsistencies be detected.

The decision marks a cautious reopening rather than a full reinstatement, reflecting Indonesia’s wider efforts to demand greater accountability from global platforms deploying advanced AI systems within its borders.

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

Education and rights central to UN AI strategy

UN experts are intensifying efforts to shape a people-first approach to AI, warning that unchecked adoption could deepen inequality and disrupt labour markets. AI offers productivity gains, but benefits must outweigh social and economic risks, the organisation says.

UN Secretary-General António Guterres has repeatedly stressed that human oversight must remain central to AI decision-making. UN efforts now focus on ethical governance, drawing on the Global Digital Compact to align AI with human rights.

Education sits at the heart of the strategy. UNESCO has warned against prioritising technology investment over teachers, arguing that AI literacy should support, not replace, human development.

Labour impacts also feature prominently, with the International Labour Organization predicting widespread job transformation rather than inevitable net losses.

Access and rights remain key concerns. The UN has cautioned that AI dominance by a small group of technology firms could widen global divides, while calling for international cooperation to regulate harmful uses, protect dignity, and ensure the technology serves society as a whole.

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

Roblox faces new dutch scrutiny under EU digital rules

Regulators in the Netherlands have opened a formal investigation into Roblox over concerns about inadequate protections for children using the popular gaming platform.

The national authority responsible for enforcing digital rules is examining whether the company has implemented the safeguards required under the Digital Services Act rather than relying solely on voluntary measures.

Officials say children may have been exposed to harmful environments, including violent or sexualised material, as well as manipulative interfaces encouraging more extended play.

The concerns intensify pressure on the EU authorities to monitor social platforms that attract younger users, even when they do not meet the threshold for huge online platforms.

Roblox says it has worked with Dutch regulators for months and recently introduced age checks for users who want to use chat. The company argues that it has invested in systems designed to reinforce privacy, security and safety features for minors.

The Dutch authority plans to conclude the investigation within a year. The outcome could include fines or broader compliance requirements and is likely to influence upcoming European rules on gaming and consumer protection, due later in the decade.

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

South Korea prepares for classroom phone ban amid disputes over rules

The East Asian country is preparing to enforce a nationwide ban on mobile phone use in classrooms, yet schools remain divided over how strictly the new rules should be applied.

A ban that takes effect in March under the revised education law, and officials have already released guidance enabling principals to warn students and restrict smart devices during lessons.

These reforms will allow devices only for limited educational purposes, emergencies or support for pupils with disabilities.

Schools may also collect and store phones under their own rules, giving administrators the authority to prohibit possession rather than merely restricting use. The ministry has ordered every principal to establish formal regulations by late August, leaving interim decisions to each school leader.

Educators in South Korea warn that inconsistent approaches are creating uncertainty. Some schools intend to collect phones in bulk, others will require students to keep devices switched off, while several remain unsure how far to go in tightening their policies.

The Korean Federation of Teachers’ Associations argues that such differences will trigger complaints from parents and pupils unless the ministry provides a unified national standard.

Surveys show wide variation in current practice, with some schools banning possession during lessons while others allow use during breaks.

Many teachers say their institutions are ready for stricter rules, yet a substantial minority report inadequate preparation. The debate highlights the difficulty of imposing uniform digital discipline across a diverse education system.

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

Google launches AlphaGenome AI tool

Google has unveiled AlphaGenome, a new AI research tool designed to analyse the human genome and uncover the genetic roots of disease. The announcement was made in Paris, where researchers described the model as a major step forward.

AlphaGenome focuses on non-coding DNA, which makes up most of the human genome and plays a key role in regulating genes. Google scientists in Paris said the system can analyse extremely long DNA sequences at high resolution.

The model was developed by Google DeepMind using public genomic datasets from humans and mice. Researchers in Paris said the tool predicts how genetic changes influence biological processes inside cells.

Independent experts in the UK welcomed the advance but urged caution. Scientists at University of Cambridge and the Francis Crick Institute noted that environmental factors still limit what AI models can explain.

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

Deezer opens AI detection tool to rivals

French streaming platform Deezer has opened access to its AI music detection tool for rival services, including Spotify. The move follows mounting concern in France and across the industry over the rapid rise of synthetic music uploads.

Deezer said around 60,000 AI-generated tracks are uploaded daily, with 13.4 million detected in 2025. In France, the company has already demonetised 85% of AI-generated streams to redirect royalties to human artists.

The tool automatically tags fully AI-generated tracks, removes them from recommendations and flags fraudulent streaming activity. Spotify, which also operates widely in France, has introduced its own measures but relies more heavily on creator disclosure.

Challenges remain for Deezer in France and beyond, as the system struggles to identify hybrid tracks mixing human and AI elements. Industry pressure continues to grow for shared standards that balance innovation, transparency and fair payment.

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

US cloud dominance sparks debate about Europe’s digital sovereignty

European technology leaders are increasingly questioning the long-held assumption that information technology operates outside politics, amid growing concerns about reliance on US cloud providers and digital infrastructure.

At HiPEAC 2026, Nextcloud chief executive Frank Karlitschek argued that software has become an instrument of power, warning that Europe’s dependence on American technology firms exposes organisations to legal uncertainty, rising costs, and geopolitical pressure.

He highlighted conflicts between EU privacy rules and US surveillance laws, predicting continued instability around cross-border data transfers and renewed risks of services becoming legally restricted.

Beyond regulation, Karlitschek pointed to monopoly power among major cloud providers, linking recent price increases to limited competition and warning that vendor lock-in strategies make switching increasingly difficult for European organisations.

He presented open-source and locally controlled cloud systems as a path toward digital sovereignty, urging stronger enforcement of EU competition rules alongside investment in decentralised, federated technology models.

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

Experts propose frameworks for trustworthy AI systems

A coalition of researchers and experts has identified future research directions aimed at enhancing AI safety, robustness and quality as systems are increasingly integrated into critical functions.

The work highlights the need for improved tools to evaluate, verify and monitor AI behaviour across diverse real-world contexts, including methods to detect harmful outputs, mitigate bias and ensure consistent performance under uncertainty.

The discussion emphasises that technical quality attributes such as reliability, explainability, fairness and alignment with human values should be core areas of focus, especially for high-stakes applications in healthcare, transport, finance and public services.

Researchers advocate for interdisciplinary approaches, combining insights from computer science, ethics, and the social sciences to address systemic risks and to design governance frameworks that balance innovation with public trust.

The article also notes emerging strategies such as formal verification techniques, benchmarks for robustness and continuous post-deployment auditing, which could help contain unintended consequences and improve the safety of AI models before and after deployment at scale.

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