Artificial intelligence Archives | Page 16 of 699

OECD maps AI and citizen participation

The OECD has published a report examining how AI could support citizen participation and democratic innovation while highlighting the safeguards needed for its responsible use.

The report, Artificial Intelligence and the Future of Citizen Participation, was approved and declassified by the OECD Public Governance Committee on 22 June 2026. It was produced as part of the OECD Public Governance Reviews series in collaboration with the Bertelsmann Stiftung.

The report says public participation can help governments design better policies and strengthen trust. It cites OECD trust findings showing that people who feel they have a say in government decisions are far more likely to report high trust in government.

The OECD notes that governments have long relied on digital technologies, including online platforms and civic tech tools, to expand public participation. AI represents the next stage of this evolution, with governments increasingly experimenting with tools for consultation, deliberation, communication and policy analysis.

The report is based on desk research and analysis of 50 AI use cases in participation processes from 22 OECD member and partner countries. It proposes a typology to help public officials and practitioners understand where AI tools may be useful and what challenges they may address.

Based on an analysis of 50 AI use cases from 22 OECD member and partner countries, the report proposes a typology covering nine categories of AI applications, including information development, sense-making, translation, transcription, virtual assistance, moderation, facilitation, simulation and participation architecture.

These tools can support both front-office activities, where citizens interact directly with government, and back-office activities, where public administrations design, analyse and manage processes internally.

According to the OECD, AI could make participation processes more accessible and efficient by helping governments analyse large volumes of public input, improve communication, reduce administrative costs and broaden participation.

Sense-making tools can help analyse large amounts of text submitted during consultations. Translation and transcription tools can make processes more accessible across languages and formats, while virtual assistants can help people navigate information about citizen participation opportunities.

AI can also support moderation and facilitation. The report says such tools may help prevent spam, hate speech or manipulation in online discussions, and could support live deliberation by identifying common ground or structuring debate.

However, the OECD cautions against treating AI as a simple fix for democratic challenges. It says technology alone cannot solve problems such as weak links between participation processes and actual policy decisions.

The report also highlights ethical, operational and societal risks, including algorithmic bias, opaque decision-making, hallucinations, cybersecurity threats, digital exclusion and declining public trust if AI systems are poorly designed or deployed.

The OECD also highlights the risks of inaction, noting that governments may miss valuable opportunities if they avoid AI tools even when they could be applied responsibly.

The report says governments should establish guardrails for AI use in citizen participation, including transparency, compliance with democratic values, protection of civic space, attention to data divides and low-tech alternatives for citizens with limited digital access.

It also calls for stronger enablers, including AI literacy, skills development, citizen engagement in the design and governance of AI systems, open standards where appropriate, and support for scaling successful pilots.

The OECD concludes that most public-sector use of AI in citizen participation remains experimental. It argues that lasting benefits will depend on transparent governance, human oversight and continued efforts to strengthen democratic participation beyond technology alone.

Why does it matter?

Governments are increasingly exploring AI as a way to make public participation more accessible, scalable and responsive. The OECD’s report shows that AI can support consultation, deliberation and policy analysis, but only when accompanied by safeguards that protect transparency, inclusion and democratic accountability.

The report also reinforces a broader shift in AI governance from technical capability to institutional design. By emphasising human oversight, civic participation, digital inclusion and democratic values, the OECD argues that AI should enhance, not replace, the processes that underpin public trust and democratic decision-making.

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

MIT develops AI system to improve robot understanding

Researchers at MIT’s Computer Science and Artificial Intelligence Laboratory have developed a system that helps robots interpret vague human instructions while using significantly less training data.

The approach, called Masked Inverse Reinforcement Learning (Masked IRL), uses two large language models to clarify tasks and identify the details that matter for safe robot movement.

One model expands ambiguous instructions based on user demonstrations. A second model filters out irrelevant information and highlights factors the robot should include in its motion plan.

The system can help robots understand unstated preferences, such as avoiding a laptop while delivering a coffee mug or keeping a safe distance from a person during a task.

MIT said Masked IRL correctly identified users’ unstated preferences up to 15% more often than comparable methods. Researchers also found that it required nearly five times less demonstration data to learn new tasks.

The approach was tested in simulated environments and on a real robotic arm. The robot completed tasks it had not seen during training, including moving a cup towards a person while avoiding a computer and handing over an object while staying away from nearby obstacles.

Researchers plan to make the system more dynamic by adding cameras, enabling robots to identify relevant objects and ignore distractions in their surroundings visually.

Why does it matter?

Masked IRL could make robots easier to deploy in homes, offices, factories and care environments by reducing the amount of human training needed. The system also addresses a core safety challenge in robotics: people often give vague instructions and leave important preferences unstated. Better interpretation of human intent could help robots work more safely around people, objects and changing environments.

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

South Korea unveils national AI infrastructure strategy

South Korea’s Ministry of Science and ICT has announced a comprehensive whole-of-government strategy to expand the country’s AI computing infrastructure and strengthen national AI capabilities.

The strategy is built around three pillars: expanding AI computing infrastructure, developing next-generation AI models, and accelerating AI adoption across public services. To strengthen computing capacity, the government aims to secure 18,000 high-performance GPUs by the first half of 2026, with 10,000 acquired through a public-private National AI Computing Centre and another 8,000 deployed as part of a sixth national supercomputer.

To advance domestic AI development, the government plans to launch a flagship initiative provisionally named the ‘World’s Best LLM’ project. Selected AI teams will receive dedicated access to computing resources, datasets and research funding. A Global AI Challenge will also be launched to attract leading domestic and international researchers, with winners offered startup support or positions within flagship AI projects.

Talent development is another key pillar. South Korea plans to expand its AI Frontier Labs beyond New York into Europe and other regions, establish AI Transformation graduate schools through industry-university partnerships and offer competitive salaries, research funding and relocation support to attract leading international AI experts.

The third pillar focuses on deploying domestically developed AI models across public services, including healthcare, education, the legal system, public administration, disaster management and content creation.

Why does it matter?

South Korea’s strategy reflects a growing global shift towards treating AI as strategic national infrastructure rather than simply a commercial technology. By combining investments in computing capacity, foundation models, talent development and public-sector deployment, the government is pursuing a comprehensive approach to strengthening technological competitiveness and digital sovereignty.

The plan also illustrates how competition in AI increasingly extends beyond model development alone. Access to high-performance computing, skilled researchers and coordinated industrial policy is becoming just as important as algorithmic innovation, with governments playing a more active role in shaping national AI ecosystems.

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

Australian audit highlights governance gaps in public-sector AI

The Australian National Audit Office has found that IP Australia’s use of AI in the patent rights process is largely effective, while calling for stronger cybersecurity governance, monitoring and strategic oversight.

Auditor-General Report No. 43 of 2025–26 examined whether IP Australia has effective arrangements to support AI adoption in the patent rights process. IP Australia administers intellectual property rights, including patents, trade marks, design rights and plant breeders’ rights.

The agency deployed its first AI tool for patent examination in 2018 and now uses four AI tools in the process. The tools are designed to provide examiners with information to support better decisions, rather than to decide patent applications themselves.

The ANAO said IP Australia has been an early adopter of AI and has progressively improved its governance arrangements. The agency has introduced an AI governance policy, risk-scaled assessment mechanisms and clearer enterprise accountability roles.

However, the audit found that strategic oversight of AI implementation and related benefits is not yet fully established. It said IP Australia’s AI inventory, committee roles and use-case ownership remain works in progress.

Monitoring and reporting were assessed as only partly effective. The ANAO said benefits have been inconsistently defined and measured, making it harder to demonstrate the ongoing effectiveness of AI tools and manage emerging risks.

The ANAO made two recommendations, urging IP Australia to review cybersecurity governance controls for AI and establish clearer risk-based monitoring and reporting arrangements. IP Australia agreed to both recommendations.

The audit said public-sector agencies should regularly reassess AI governance frameworks as they move from experimentation to wider use.

Why does it matter?

The audit shows how AI is moving from experimentation into routine public-sector decision support. IP Australia’s experience points to the benefits of AI in improving efficiency and quality, but also shows that governance must evolve as tools become embedded in official processes. Cybersecurity, accountability, monitoring and measurable benefits are becoming central to responsible AI use in government.

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

HP adopts OpenAI Frontier for enterprise AI

HP has announced a strategic partnership with OpenAI to integrate OpenAI Frontier across parts of its customer-facing services and internal operations.

The company said the partnership supports its broader future-of-work strategy, with an initial focus on customer experience, partner services, employee productivity and software development.

HP plans to use OpenAI Frontier to create a more consistent experience across its retail, partner, chat and voice channels, helping customers and partners resolve routine queries and complete workflows more efficiently.

The company said it is among the first global enterprises to adopt the Frontier platform. While specific use cases will evolve over time, the initial focus includes customer and partner tools, telemetry insights through the Workforce Experience Platform, employee productivity and software development.

These include customer and partner-facing tools, customer telemetry insights and reporting through HP’s Workforce Experience Platform, employee productivity and software development.

The partnership follows an evaluation phase that began in February 2026, during which HP assessed OpenAI Frontier’s technical capabilities, enterprise integration, security features and potential business applications.

HP said it also tested agentic capabilities during the evaluation. The company said the results led it to conclude that OpenAI offers models and agent-based capabilities aligned with its strategic priorities.

HP and OpenAI now plan to co-develop future use cases. HP said these will need to meet its enterprise standards, particularly around data integration, governance and security.

OpenAI said HP had already used OpenAI APIs and tools such as ChatGPT and Codex in early projects. The companies said the new partnership is intended to move beyond pilots toward broader enterprise deployment.

HP also linked the partnership to its broader AI hardware strategy, saying it is developing agentic AI devices and dedicated hardware designed to support continuous AI inference and integrate seamlessly into workplace workflows.

For AI workloads that require continuous inference, HP said it is building devices with dedicated hardware optimised to run agentic AI workloads around the clock.

HP also pointed to its Workforce Experience Platform, which is used to manage device fleets and provide telemetry insights across PCs, workstations, printers and collaboration tools.

HP said the partnership reflects a broader shift from isolated AI pilots to enterprise-wide deployment, with AI increasingly serving as an operating layer embedded across customer services, software development and business operations.

Why does it matter?

The partnership illustrates how large enterprises are moving beyond experimental AI deployments towards organisation-wide integration. Rather than treating AI as a standalone application, companies are increasingly embedding it into customer support, software development, employee productivity and operational workflows.

It also highlights the growing importance of enterprise AI governance. As organisations deploy increasingly capable agentic systems, success will depend not only on model performance but also on secure integration, data governance and oversight that ensure AI can operate reliably within existing business processes.

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

South Korea plans $518 billion semiconductor hub for AI demand

Samsung Electronics and SK Hynix have announced plans to invest a combined 800 trillion won, about $518 billion, in a new semiconductor manufacturing hub in South Korea’s southwest.

The two companies, which together produce around two-thirds of the world’s memory chips, will each build two new fabrication plants outside their existing manufacturing base in Gyeonggi Province.

Samsung’s new facilities are planned for the city of Gwangju, with several possible sites under consideration, including land linked to a military air base planned for relocation.

The investment responds to rising demand for memory chips used in AI data centres, industrial robotics and autonomous vehicles. Existing semiconductor facilities in Gyeonggi Province are expected to face capacity pressure sooner than previously projected.

South Korea’s government is also linking the project to a broader strategy to build a nationwide semiconductor ecosystem. Existing hubs in the Southeast are expected to expand chip component and material production. At the same time, the central Chungcheong region will focus on chip packaging, and data centres will be developed across the country.

The project also supports the government’s goal of spreading major technology investment beyond the Seoul metropolitan area, where much of the country’s semiconductor industry has historically been concentrated.

Why does it matter?

The planned investment shows how AI demand is driving long-term semiconductor capacity expansion at a national scale. Memory chips are central to AI data centres and high-performance computing, and Samsung and SK Hynix remain two of the most important suppliers in the global market. South Korea’s decision to link new chip fabrication with regional development also shows how AI infrastructure is becoming part of broader industrial and economic planning, not only technology strategy.

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

UNDP outlines responsible AI use in electoral administration

UNDP and the UN Department of Political and Peacebuilding Affairs (DPPA) have published a technical guide on AI in electoral administration to help election authorities assess the responsible adoption of the technology.

The publication, From Promise to Practice: AI in Electoral Administration, was produced jointly by UNDP and DPPA’s Electoral Assistance Division. It is intended as a practical resource for electoral management bodies considering how AI could support their work.

The report notes that AI is not new to elections, with electoral authorities already using technologies such as biometric voter identification, optical ballot scanning and algorithmic analysis of voter registration databases.

However, the report argues that generative AI, large language models and agentic systems represent a significant shift. While they could improve public outreach, anomaly detection, organisational efficiency and voter communication, they also introduce risks related to hallucinations, bias, reliability and public trust.

The publication stresses that AI adoption should form part of a broader digital transformation strategy, including stronger data governance, digital public infrastructure and organisational capacity.

The report says effective AI adoption in elections depends on political and public consensus, reliable technical implementation, and transparent governance. It notes that past use of digital technologies in elections shows the need for clear problem definition, rigorous testing, and gradual adoption.

Reliability is identified as a central concern. The report warns that inaccurate or misleading AI outputs could affect voter information, election operations and public confidence. In electoral settings, even minor errors can affect voting rights, trigger legal disputes or undermine trust in election outcomes.

The publication also highlights inclusion as a core requirement. AI systems can support inclusion if they are designed with representative data, inclusive testing, participatory design, and continuous monitoring. However, biased datasets or poorly designed systems can disadvantage women, young people, persons with disabilities, minorities, and other groups.

Data governance is another major theme. Electoral management bodies often hold sensitive personal data, including biometric information, while operating under strong transparency expectations. The report says principles such as proportionality, informed consent, and data quality must be translated into practical policies.

The report groups AI applications into five functional areas: analysis, recognition, automation, content creation and voter communication. Examples include anomaly detection, biometric verification, workflow automation, multilingual outreach and AI-powered chatbots.

The publication identifies 12 features to guide electoral management bodies. The features include understanding the need, building political consensus, protecting rights, managing risk, ensuring human oversight, testing early and often, designing for inclusion, forming skilled and diverse teams, building securely, addressing privacy, defaulting to open approaches where appropriate, and designing systems for the future.

The report also links AI in electoral administration to the Global Digital Compact, which promotes a responsible, transparent, accountable, and human-centric approach to emerging technologies. It says electoral authorities should consider how commitments on digital public infrastructure, open-source tools, safeguards, data standards, and human oversight apply to their work.

UNDP and DPPA say the value of AI in elections should be measured by whether it makes electoral processes more credible, inclusive, and resilient, as well as more efficient.

Rather than endorsing AI for electoral processes, the publication provides a framework to help electoral authorities assess whether, where and how AI can be adopted responsibly.

Why does it matter?

Elections are among the most sensitive public processes, meaning AI systems must be deployed with exceptional care. While AI could improve administrative efficiency, voter communication and fraud detection, failures involving accuracy, bias, privacy or transparency could undermine public confidence and the integrity of electoral processes.

The guidance also reflects a broader shift in AI governance from high-level principles to practical implementation. By focusing on human oversight, data governance, inclusion, testing and institutional capacity, UNDP and DPPA are encouraging election authorities to treat AI as a governance challenge that requires careful planning rather than a simple technological upgrade.

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

UNESCO advances AI ethics training in Mexico’s judiciary

UNESCO has delivered the first specialised in-person training programme on the ethical use of AI for judicial professionals in Mexico City, aiming to support the responsible adoption of AI across the country’s justice system.

More than 50 civic judges, mediators and public defenders took part in the programme, which focused on ensuring AI supports judicial processes in Mexico while respecting transparency, accountability and human rights.

The programme introduced participants to the opportunities and risks associated with AI in judicial decision-making while providing practical guidance on applying ethical safeguards in courts and public institutions.

The training was based on UNESCO’s Recommendation on the Ethics of Artificial Intelligence, adopted by all UNESCO Member States in 2021, and incorporated the organisation’s newly published Guidelines for the Use of AI Systems in Courts and Tribunals.

The initiative forms part of a broader UNESCO and European Commission project supporting countries in implementing AI governance frameworks through capacity building, technical assistance and policy tools.

Participants were also introduced to UNESCO’s practical governance tools, including the Readiness Assessment Methodology, the Ethical Impact Assessment framework and Global Toolkit on AI and the Rule of Law for the Judiciary.

UNESCO emphasised that although AI is increasingly being incorporated into judicial and administrative processes, human oversight must remain central. The organisation said well-trained judicial professionals are essential to ensuring AI improves access to justice without replacing human judgement or undermining fundamental rights.

Why does it matter?

As AI becomes more common in courts and public administration, effective governance depends not only on regulation but also on the ability of judges and other legal professionals to understand the technology’s capabilities, limitations and risks. Training programmes such as this can help ensure AI supports judicial work without compromising due process, transparency or fundamental rights.

The initiative also demonstrates UNESCO’s broader approach to AI governance, combining international ethical principles with practical implementation tools. By equipping judicial institutions with guidance, assessment frameworks and technical expertise, the organisation aims to help countries translate high-level AI principles into everyday public-sector practice.

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

Australia pushes more AI nudify services offline over child safety

Three more AI-powered ‘nudify’ services have withdrawn access for Australian users after enforcement action by Australia’s eSafety Commissioner under the country’s Age-Restricted Material codes.

The codes require AI services that allow users to access or generate age-restricted material, including sexually explicit material, to put appropriate age-assurance measures in place to prevent access by children under 18.

The latest action followed a formal Direction to Comply issued to one of the most widely used nudify services in Australia, requiring the provider to implement stronger protections within 14 days. Instead, the company disabled access for Australian users, while two associated services also withdrew.

eSafety said users in Australia will no longer be able to log in or use the service’s features, although landing pages may remain visible with content blurred.

The regulator said AI nudification tools pose serious risks because they can be used to create non-consensual sexually explicit deepfakes and child sexual exploitation material. It has also warned that such tools are increasingly being misused in school settings.

The action is part of eSafety’s broader enforcement focus on generative AI and nudify services now that Australia’s online safety codes and standards are in force. The regulator said seven of the most frequently accessed nudify services in Australia have either withdrawn from the market or introduced age-assurance measures following intervention.

Australia is also preparing further legislation to prohibit nudify services used to generate non-consensual sexually explicit material.

Why does it matter?

Australia’s approach shows how regulators can use age-assurance and online safety rules to restrict children’s access to high-risk generative AI tools before new AI-specific laws are fully in place. The case is also important because nudify services sit at the intersection of AI-generated abuse, child protection, image-based harm and platform accountability. By forcing services to either introduce safeguards or withdraw access, eSafety is creating a practical enforcement model that other jurisdictions may closely watch.

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

OpenAI previews GPT-5.6 Sol model with stronger safeguards

OpenAI has begun a limited preview of GPT-5.6 Sol, a new flagship model in its new GPT-5.6 family, which also includes Terra and Luna. The company said all three models are expected to become generally available in the coming weeks.

The company said the preview is initially limited to a small group of trusted partners. OpenAI said it shared its release plans and model capabilities with the U S government before launch and is initially limiting access at the government’s request.

The company said it does not consider government pre-release access an appropriate long-term default. Instead, it described the limited preview as a temporary measure while working with the US administration on a repeatable release framework linked to a cybersecurity Executive Order.

OpenAI described GPT-5.6 Sol as its most capable model to date, highlighting improvements in agentic coding, biology and cybersecurity while saying a broader set of evaluation results will be published when the model becomes generally available.

For coding, OpenAI said GPT-5.6 Sol set a new state of the art on Terminal-Bench 2.1, which tests command-line workflows involving planning, iteration and tool coordination.

The company also reported improvements in biology workflows. On GeneBench v1, which evaluates long-horizon genomics and quantitative biology tasks, OpenAI said the model performed better than GPT-5.5 while using fewer tokens.

Cybersecurity is a major focus of the preview. OpenAI said GPT-5.6 Sol is its most capable model yet for cybersecurity tasks, including vulnerability research and exploitation-related workflows.

OpenAI said the model performs better at identifying and helping remediate vulnerabilities than at carrying out end-to-end offensive cyber operations. According to the company, GPT-5.6 Sol did not exceed the Cyber Critical threshold under its Preparedness Framework.

OpenAI said the GPT-5.6 release includes its most robust safeguards to date, with configurations tailored to each model’s capabilities. The company said these safeguards are intended to constrain prohibited offensive use while preserving access for legitimate work such as code review, vulnerability research, patch development, debugging, security education and defensive testing.

Safeguards include model-level protections, real-time generation checks, account-level monitoring, differentiated access controls, enforcement mechanisms and ongoing testing. OpenAI said some higher-risk requests may be delayed or blocked during the preview period.

The company said it devoted more than 700,000 A100-equivalent GPU hours to automated red-teaming, complemented by third-party expert testing, to evaluate the model’s resilience against jailbreak attempts.

During the preview, GPT-5.6 models will initially be available through the API and Codex to selected trusted partners and organisations. OpenAI said broader access for ChatGPT, Codex and API users is planned soon.

During the preview, GPT-5.6 models will be available through the API and Codex to selected partners. OpenAI said broader access across ChatGPT, Codex and the API is planned soon. It also announced pricing for the model family and said GPT-5.6 Sol will launch on Cerebras in July, initially for a limited group of customers.

Why does it matter?

GPT-5.6 Sol illustrates how frontier AI releases are becoming increasingly governed by phased deployment, targeted access and extensive safety testing rather than immediate public availability. OpenAI’s emphasis on cybersecurity evaluations, automated red-teaming and layered safeguards reflects growing efforts to manage the risks associated with increasingly capable foundation models.

The rollout also highlights the evolving relationship between AI companies and governments. By combining limited pre-release access, enterprise deployment and structured safety frameworks, OpenAI is helping shape emerging norms for how advanced AI systems are evaluated, governed and introduced into real-world use.

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