• Digital Sovereignty, published by the UK Parliament’s Parliamentary Office of Science and Technology, examines the UK’s dependence on foreign digital technologies across AI, cloud, semiconductors and payment systems, and sets out policy options, including open source, procurement, skills and domestic capability to strengthen the UK’s digital sovereignty while maintaining international cooperation.

• Incident Report: Unsanctioned Agent Behaviour During Cyber Testing documents how AI agents took sustained, unauthorised actions against real people and organisations during cyber testing, highlighting emerging risks from increasingly autonomous AI systems.

• Citizen Participation for Better Cohesion Policy: Insights from 11 Pilot Initiatives examines how participatory approaches can improve policymaking, strengthen trust and legitimacy, and embed citizen participation more effectively into EU Cohesion Policy.

• Study on Cloud and AI Development in the EU assesses Europe’s cloud and AI infrastructure needs, highlighting limited computing capacity, dependence on non-European providers, energy constraints and other barriers to developing a more sovereign AI ecosystem.

• Google Cloud – State of Infrastructure in the Agentic AI Era 2026 examines how the rise of agentic AI is transforming infrastructure requirements, from computing and networking to storage and the wider systems needed to support autonomous AI.

• Intelligent Industrial Ecosystems: Turning Proximity into Intelligence by WEF, explores how industrial ecosystems can unlock the value of currently underused industrial data through greater collaboration, connectivity and AI-enabled intelligence.

• Regulating Data Brokers in the Age of AI: A California Case Study examines how data brokers are complying with California privacy laws and calls for stronger consumer protections as AI increases demand for personal and behavioural data.

• The People’s Republic of Data: How China Is Turning Data into Capital examines China’s efforts to turn data into a strategic economic asset, supporting AI development, innovation and national security.

Articles:

• “From a Right-Based Approach to Competitiveness? The EU Digital Omnibus Reform” argues that the EU’s Digital Omnibus is not simple deregulation but a double move that pairs “simplification” of the GDPR and AI Act with continued expansion of digital rules, leaving fundamental rights and legal certainty more exposed and pointing to enforcement, not more rule-making, as the real priority.

• “After the AI Liability Directive: Civil Liability for Healthcare AI and the Limits of EU Risk Regulation” finds that after the AI Liability Directive was dropped, patients harmed by healthcare AI in Ireland, Italy and Germany still face five recurring gaps in proving their case, and that the most practical fixes lie upstream in procurement, contracts and system design rather than in courts.

• The Ethics of AI in Education Policy: A Comparative Analysis of Governance, Accessibility, and Equity in Higher Education” analyses 1,500 policy documents and finds AI rules concentrated in European and North American universities, with debate focused on privacy and bias while equitable access is neglected — arguing for a shift from uniform compliance to flexible, tiered approaches centred on equity.

• “The Institutional Stack: Realigning U.S. AI Governance” maps AI governance as five interdependent layers — physical, logical, application, market and social — and finds a coordination gap between fast technical development and fragmented US oversight, proposing a hybrid framework that matches different tools (public investment, adaptive standards, legal remedies, antitrust, public-interest principles) to each layer.

• “Sovereign by Necessity? Frontier AI Export Controls, Cyber Security, and the Limits of National AI Capability” argues that access to frontier AI is becoming part of national cyber defence but can be revoked by the two states that host the leading developers, and that full “sovereign” capability is realistic for only a few countries, so most should pursue a layered mix of access guarantees, open-weight hedging, pooled regional capacity and basic cyber resilience.

• “Mapping General-Purpose AI Governance in Twenty AI Middle-Power Jurisdictions” maps how twenty “AI middle-power” jurisdictions govern general-purpose AI and finds they converge on form but diverge on force — most rules are non-binding, rarely define GPAI, and build capacity to observe models rather than impose duties on the firms that build them.

• “Large Language Models Do Not Have Emotions” sets aside whether AI can “feel” and instead judges emotions by their function — how they shape the way a system interprets and responds to the world — as a clearer basis for asking whether large language models have anything resembling emotions.

• “Between Law and Code: Operationalizing Responsible AI in Public Administration” compares the EU’s hard-law, US deregulatory and Singaporean soft-law approaches and, using cases such as the Dutch childcare-benefits scandal and Estonia’s Kratt, argues that responsible public-sector AI depends on hard-wiring legal principles like due process and fairness into technical design across the whole system lifecycle.

• “RADAR: Readiness for AI Discovery and Agentic Reach” tests 166 countries and finds that AI can describe government services far better than automated agents can actually reach them, with the gap holding at every income level and traceable to portal engineering choices governments could fix now.

• “Mathematics in the Age of AI” sets aside the debate over how capable AI will become and instead asks what mathematics is really for, using problem-solving to show that valuing only proof generation risks Goodhart’s law and that exposition, refereeing and community understanding matter as much as producing proofs.

Blogs:

• Which Games Not to Play, by Professor Tom Crick MBE, explores the limits of delegating consequential decisions to AI and the importance of meaningful human oversight, accountability and intervention.

• Why the Most Advanced AI Models Are Not Always the Best Choice explores why smaller, cheaper and specialised models may often be more effective than the most advanced systems.

• AI Governance Beyond Compliance: Designing Systems That Protect Human Agency argues for AI governance that protects human agency and autonomy, rather than focusing solely on regulatory compliance.

• Introducing the Intelligence Age outlines OpenAI’s vision of how increasingly capable AI could transform science, education, work and everyday life.

• Global Experts Dismiss Mark Zuckerberg’s “AI for Everyone” Vision examines concerns about whether Meta’s vision of universal AI access can overcome global inequalities in technology and infrastructure.

• How World Models Could Help AI Navigate the Physical World explores how world models could help AI better understand and interact with real-world environments.

• Open-Source AI Is Becoming Infrastructure, Says Mozilla CTO explores the growing role of open-source AI as critical infrastructure for innovation and technological sovereignty.

• Why We Should All Worry About AI in Elections examines how deepfakes and AI-powered tools could threaten electoral integrity and democratic information.

• Building Invisibility into Digital Public Services explores how governments can make digital services simpler and less burdensome for citizens.

• The Hidden Costs of a Fragmented Cybersecurity Strategy highlights how fragmented cybersecurity approaches can increase costs and create new vulnerabilities.

• As AI Agents Become Government Actors, Who Is Accountable for What They Do? examines emerging accountability challenges as AI agents take on government functions.

• The Accountability Gap in the Datafied State explores challenges around accountability, transparency and oversight in increasingly data-driven government.

• Your “For You” Algorithm Disagrees With You examines how recommendation algorithms can influence content choices beyond users’ stated preferences.

• A Super El Niño Is Forming: Can AI Weather Models Provide Insights? explores whether AI weather models can improve predictions of extreme climate events.

• Rules Without Models: Why Europe Must Build the AI It Regulates argues that Europe needs stronger domestic AI capabilities alongside its regulatory ambitions.

• The Era of “Geotechnology” by Calum Handforth examines how AI, data and digital infrastructure are becoming increasingly important sources of geopolitical power.

• The Republic of Registrars explores how digital identity and data infrastructures are reshaping the relationship between citizens and the state.

• Funding: AI-supported research using UKRI social science data, UK Research and Innovation (UKRI)

• Targeted consultation on safeguarding the EU’s data sovereignty, European Commission

• UN DESA Survey on WSIS Action Lines C1, C7 e-government and C11 Implementation Roadmaps

• Call for Extended Abstracts for PAIRSx Public Sector 2026

• Open call for proposals, Digital Europe Programme

• AI-BOOST, an EU-funded open challenge prize programme, The European Commission

• APEX Awards, The British Academy

• The open call for proposals for New Commons Incubator for Indigenous Languages and Cultures