EU court annuls Meta Marketplace designation

The General Court of the European Union has annulled the European Commission’s decision designating Meta as a gatekeeper for Marketplace under the Digital Markets Act, while upholding the company’s designation for Messenger.

The case concerned the Commission’s 5 September 2023 decision designating Meta as a gatekeeper for several core platform services, including Facebook, Messenger, and Marketplace. Meta challenged the decision in part, contesting the classification of Messenger and Marketplace as important gateways under the DMA.

The General Court upheld the Commission’s assessment of Messenger, finding that the service is a number-independent interpersonal communications service distinct from Facebook. The court said Messenger is available through standalone applications, can be used independently of Facebook, and includes tools that allow businesses to engage with users.

The court also found that the Commission did not have to count only Messenger users who were not also Facebook users when assessing whether the quantitative threshold under the DMA was met. It also said the Commission was not required to open a market investigation in the absence of sufficiently substantiated arguments from Meta calling the DMA presumptions into question.

For Marketplace, the court found that the Commission erred in law by relying only on data from the three years preceding designation without taking account of changes made at the end of July 2023. Those changes limited the number of listings that could be published per user and led to the disappearance of the criterion used by the Commission to identify business users.

The court also found that the Commission had not provided sufficient reasoning for classifying Marketplace as an online intermediation service. It said the Commission failed to provide a concrete analysis of the July 2023 changes or to explain their effect on whether Marketplace-enabled business users could offer goods and services to consumers.

As a result, the decision was annulled only to the extent that it designated Meta as a gatekeeper for Marketplace. Meta’s Messenger designation remains in place.

Why does it matter?

The judgement is an important test of how the EU courts will review Digital Markets Act gatekeeper designations. It confirms that the Commission can rely on DMA presumptions where companies do not provide sufficiently substantiated counterarguments, as seen with Messenger. But it also shows that the Commission must properly assess relevant changes and provide sufficient reasoning when classifying a service as a core platform service, as the Marketplace annulment demonstrates.

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

MIT develops ChartNet dataset to improve AI chart understanding

MIT researchers have developed a new dataset, ChartNet, to improve how vision-language models interpret charts and other graphical data.

The dataset is designed to help AI systems better combine visual, numerical, and linguistic information, a task that remains difficult even for advanced models. MIT said chart understanding is important for applications such as business trend analysis, financial reporting, and scientific figure interpretation.

ChartNet contains more than one million synthetic chart images, each paired with supporting code, numerical tables, textual descriptions, and question-and-answer pairs. The dataset was created through an automated pipeline that generates and augments chart examples, supported by quality checks to ensure that the code is executable and the resulting charts are accurate and clean.

The researchers developed ChartNet to address a key limitation in current AI systems: the lack of large, high-quality training data for robust chart interpretation. Many existing datasets rely on limited chart images collected from the internet and lack the supporting information needed for models to understand the underlying data.

MIT researchers used ChartNet to train several open-source vision-language models, including IBM’s Granite Vision series. The dataset improved model accuracy across chart reconstruction, chart data extraction, chart summarisation, and chart question answering.

In MIT’s testing, smaller open-source models trained with ChartNet consistently outperformed much larger commercial models on several chart-interpretation tasks. The researchers said the dataset could help smaller organisations use AI for analytical work without relying only on large proprietary systems.

Why does it matter?

ChartNet shows how better training data can improve AI performance in specialised analytical tasks. If smaller open-source models can interpret charts more accurately after training on high-quality datasets, organisations with limited budgets may gain access to stronger AI tools for business analytics, research, financial reporting, and scientific communication. The work also highlights a broader point in AI development: model capability depends not only on size, but also on the quality and structure of training data.

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

UK CMA targets AI search content use in new Google conduct requirements

The UK’s Competition and Markets Authority (CMA) has imposed a new conduct requirement on Google Search under the country’s digital markets competition regime. The measure is designed to give publishers greater control over how their content is used and to improve transparency for users.

Under the new requirement, publishers will be able to prevent their content from being used in Google’s AI-powered search features, including AI Overviews. The CMA said the measure is intended to strengthen publishers’ ability to negotiate content licensing and usage agreements with Google.

Google will also be required to provide clearer attribution for publisher content used in AI-generated search results through prominently visible links. Following consultation feedback, publishers will also be able to opt out of having their content used to fine-tune Google’s AI models.

The CMA said it will continue monitoring Google’s AI-related changes to search and may introduce additional measures if competition concerns persist. Google will have up to nine months to implement the requirements and must publish regular compliance reports as the rollout progresses in the UK.

Why does it matter?

The decision highlights growing regulatory scrutiny of how AI-powered search systems use third-party content. As search engines increasingly generate answers directly within search results, publishers have raised concerns about attribution, traffic losses and the use of their content for AI training.

The UK’s approach could influence broader debates about the relationship between AI platforms, publishers and competition policy, particularly as regulators seek to balance innovation with transparency and fair commercial practices.

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

NVIDIA expands global AI Cloud network to support sovereign and agentic AI

NVIDIA has announced a major expansion of its AI Cloud ecosystem, supporting the rapid global deployment of AI factory infrastructure designed to meet growing demand for agentic AI, physical AI, sovereign AI and large-scale inference workloads.

The initiative aims to expand access to high-performance computing resources for enterprises, startups, governments, researchers and AI developers worldwide.

According to NVIDIA, the ecosystem now spans six continents, with new partners expanding AI Cloud infrastructure across multiple regions. The company said the expansion is intended to bring AI computing resources closer to users, industries and national AI initiatives while supporting regional and sovereign AI requirements.

Several cloud providers are expanding infrastructure to support advanced AI applications, including model training, fine-tuning, inference and AI agent development. Companies including CoreWeave, Firmus, Nebius and others are deploying new AI factories capable of supporting model training, fine-tuning, inference and AI agent development.

The expansion also includes support for emerging physical AI and robotics workloads through platforms such as NVIDIA Cosmos and Isaac.

NVIDIA also highlighted growing adoption of its DSX platform, which is designed to help cloud providers deploy and manage AI factories more efficiently. The company said AI infrastructure is increasingly being assessed using metrics such as cost per token, energy efficiency and infrastructure utilisation, rather than raw computing capacity alone.

Why does it matter?

The expansion highlights the growing importance of AI infrastructure as governments and companies compete to secure the computing resources needed for advanced AI systems. Access to large-scale computing capacity is increasingly viewed as a strategic asset, particularly as countries pursue sovereign AI initiatives and seek greater control over critical digital infrastructure.

The announcement also reflects a broader shift in the AI industry, where demand is expanding beyond model training to include inference, autonomous agents and robotics applications, placing new emphasis on infrastructure efficiency, energy use and geographic distribution of computing resources.

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

France secures a €75 billion SoftBank investment for AI data centres

SoftBank Group has announced plans to develop and operate 5 GW of AI data centre capacity in France, representing an investment of up to €75 billion.

The commitment was announced at the 2026 Choose France summit and marks SoftBank Group’s largest AI infrastructure investment in Europe. The company said the project is designed to expand access to high-performance computing capacity and strengthen France’s role as a European hub for AI infrastructure.

The first phase includes an initial €45 billion investment to deliver 3.1 GW of AI data centre capacity in the Hauts-de-France region by 2031. Planned sites include Dunkirk, Bosquel, and Bouchain, with additional projects expected elsewhere in France.

The infrastructure is intended to support demand for high-performance computing from AI companies, cloud providers, enterprises, public institutions, and research organisations.

A major component of the initiative is a strategic industrial partnership with Schneider Electric. The companies will establish a large-scale industrial production cluster at the Port of Dunkirk focused on data centre infrastructure.

The cluster will include two facilities: one operated by SoftBank Group to manufacture enclosures, and one operated by Schneider Electric to integrate data centre power modules. The partnership will combine SoftBank’s robotics and automation capabilities with Schneider Electric’s energy technology expertise and local supply chain network.

SoftBank said the project is expected to create thousands of high-skilled jobs across data centre development, engineering, energy systems, robotics, operations, maintenance, and advanced manufacturing. The company also plans to support regional research and development through partnerships with universities, engineering schools, and training institutions.

Why does it matter?

SoftBank’s project would significantly expand Europe’s AI compute capacity at a time when data centres, energy infrastructure, and advanced manufacturing are becoming central to AI competitiveness. The investment also links digital sovereignty with industrial policy: France is not only seeking more AI computing infrastructure, but also a localised supply chain for data centre equipment, power systems, robotics, and technical skills.

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

Meta turns to subscriptions amid growing AI infrastructure costs

Meta has launched paid subscription plans for Facebook, Instagram and WhatsApp as part of a broader effort to diversify revenue beyond advertising. The new offerings form part of a subscription strategy called ‘Meta One‘.

Meta said the subscriptions include additional features for users, while separate premium offerings for creators, businesses and Meta AI users are currently being tested. The company indicated that these future services will also sit under the Meta One umbrella.

The announcement comes as Meta continues to increase spending on AI infrastructure. The company has projected capital expenditure of between USD 125 billion and USD 145 billion in 2026, much of it linked to AI data centres, increasing investor attention on how those investments will generate returns.

According to Euronews, Meta shares rose following the announcement. The company said subscription products will roll out globally, while some future Meta One offerings are expected to begin testing in selected markets outside the EU.

Why does it matter?

The launch of Meta One marks a further shift in Meta’s business strategy as the company looks to diversify revenue beyond digital advertising. Subscription services could provide new income streams while supporting investments in AI infrastructure and premium digital products.

The move also reflects a broader trend among technology companies seeking alternative business models as competition intensifies and AI development costs continue to rise.

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

Anthropic launches Claude Opus 4.8 with improved reasoning capabilities

Anthropic has introduced Claude Opus 4.8, an upgraded version of its flagship AI model, with improvements across coding, agentic tasks, reasoning, and practical knowledge work.

The company said the model builds on Claude Opus 4.7 and is available at the same regular pricing. Anthropic also said that fast mode for Opus 4.8 can run 2.5 times as fast and is now 3 times cheaper than fast mode for previous models.

A key focus of the release is reliability. Anthropic said early testers found Opus 4.8 sharper in judgement when performing agentic tasks, more likely to flag uncertainty, and less likely to make unsupported claims. The company’s evaluations also found the model to be around four times less likely than its predecessor to leave flaws in its own code unremarked.

New features include dynamic workflows in Claude Code, available in research preview, allowing Claude to plan and run hundreds of parallel subagents in a single session for large-scale tasks. Anthropic said the feature can support codebase-scale migrations across hundreds of thousands of lines of code.

Users on claude.ai and Claude Cowork can also control how much effort Claude applies to a response. Higher effort settings are designed to improve quality for difficult tasks, while lower effort settings allow faster responses and slower use of rate limits.

Anthropic also reported stronger alignment results for Opus 4.8 compared with Opus 4.7. Its alignment assessment found lower rates of misaligned behaviour, such as deception or misuse of cooperation, and stronger support for user autonomy and user interests.

The model is available across Anthropic’s platforms, and developers can access it through the Claude API using the claude-opus-4-8 model name.

Why does it matter?

Claude Opus 4.8 shows how frontier AI competition is moving beyond benchmark performance towards reliability in professional workflows. Features such as effort control, dynamic workflows, cheaper fast mode, and stronger agentic task performance point to a market shift in which AI systems are expected to manage longer, more complex work in coding, research, analysis, and enterprise operations while giving users greater control over cost, speed, and reasoning depth.

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

European Commission fines Temu €200 million under DSA

The European Commission has imposed a €200 million fine on Temu after finding that the online marketplace breached obligations under the Digital Services Act by failing to properly assess and mitigate systemic risks linked to illegal products sold to consumers in the EU.

According to the Commission, Temu’s 2024 risk assessment did not meet DSA requirements because it relied on general information about the wider e-commerce sector rather than evidence specific to its own platform. Regulators also found that the company significantly underestimated the likelihood that the EU consumers would encounter illegal or unsafe products.

The investigation drew on mystery shopping exercises and information from customs and market surveillance authorities. Findings included chargers that failed basic safety requirements and baby toys that contained chemicals above legal limits or presented choking hazards.

Regulators also criticised Temu for failing to sufficiently assess how recommender systems and influencer promotion programmes could contribute to the spread of illegal products on the platform.

Temu must now submit a detailed action plan explaining how it will address the shortcomings identified by the Commission. The plan will be reviewed with the European Board for Digital Services before implementation requirements are set. Failure to comply could lead to additional penalties under the DSA.

The decision is part of a wider Commission investigation into Temu, including issues related to potentially addictive design, recommender systems, and data access for researchers.

Why does it matter?

The fine marks one of the most significant enforcement actions under the Digital Services Act against a major online marketplace. It shows that the DSA is being used not only to address illegal content, but also to require platforms to assess and reduce consumer safety risks linked to illegal and unsafe goods. The case reinforces the EU’s focus on proactive risk management by very large online platforms, including how marketplace design, recommendations, and influencer promotion can amplify the reach of harmful products.

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

ECB explores AI tools for monitoring financial stability risks

The European Central Bank (ECB) has examined how AI could support financial stability monitoring and communication, comparing traditional dictionary-based sentiment analysis with transformer models and GPT-based systems. The study was published as part of the ECB’s May 2026 Financial Stability Review.

Researchers analysed all ECB Financial Stability Review publications between 2004 and 2025 to evaluate how AI systems interpret financial stability risks and vulnerabilities. The study found that GPT-based models were better able to isolate explicit risk assessments and identify stronger signals during periods of financial stress, including the global financial crisis and the COVID-19 pandemic.

The ECB also introduced its SPOT indicator, an AI-powered system that uses large language models and financial news coverage to assess the severity and probability of potential financial stability triggers. According to the study, the system detected elevated risk levels ahead of several major geopolitical and economic disruptions.

Despite the growing capabilities of AI-based analysis, the ECB stressed that such tools should remain complementary to human expertise, vulnerability analysis and stress testing. The ECB stressed that financial stability assessments cannot rely solely on automated systems because forecasting shocks and systemic crises remains inherently uncertain.

Why does it matter?

Central banks are increasingly exploring AI tools to analyse large volumes of financial reports, market data, and news coverage. The ECB’s findings suggest that advanced AI models could help identify emerging vulnerabilities and support risk monitoring, while also highlighting the continued need for human judgement in assessing complex financial and geopolitical developments.

As financial systems become more interconnected, AI may become an increasingly important component of central bank analytical toolkits.

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

European digital identity wallets and digital sovereignty discussed at EuroDIG 2026

Participants at EuroDIG 2026 discussed whether the European Digital Identity Wallet (EUDI Wallet) and the proposed EU business wallet could become foundational infrastructure for a more integrated European digital single market.

The session focused on reducing fragmentation in digital identity, trust, and cross-border administrative procedures across Europe. Speakers broadly supported the wallets as tools for simplification and interoperability, while also raising concerns about adoption, implementation, and Europe’s dependence on non-European digital infrastructure.

Nathan Meurens, CTO of EURid, framed the discussion around a central challenge for European integration: despite advances in the single market, digital trust and identity systems remain fragmented across member states.

Norbert Sagstetter, Head of Unit for Digital Identity and Trust at the European Commission, outlined the timeline and objectives of the updated European Digital Identity Framework. He said all EU member states will be required to offer a voluntary European Digital Identity Wallet by the end of 2026 under common technical and legal standards.

According to Sagstetter, the wallet is designed to allow citizens to identify themselves digitally, sign documents, and share verified attributes and official records under user control across both public and private services.

He described the wallet as a response to the growing role of private technology platforms in digital identity management. The framework, he said, aims to provide a citizen-controlled alternative governed by European law and supervised by public authorities.

Sagstetter also highlighted the concept of ‘electronic attestation of attributes,’ which could include digital versions of documents such as driving licences, educational credentials, certificates, and potentially health-related records.

The discussion also focused heavily on the proposed EU business wallet, which remains under negotiation.

Sagstetter argued that companies across Europe still face fragmented identification systems and inconsistent administrative procedures, particularly in cross-border interactions. The business wallet, he said, is intended to enable legally effective exchange of documents and interoperable communication between companies and public administrations.

Benjamin Knirsch of SMEunited said small and medium-sized enterprises continue to face duplicated reporting obligations, fragmented platforms, and repeated verification procedures across Europe. He argued that reusable verified company information and implementation of the once-only principle could significantly reduce administrative burdens for SMEs.

However, Knirsch warned that simplification would fail if interoperability remained partial or uneven. He criticised proposals that would exempt smaller municipalities from participation, arguing that SMEs often interact most frequently with local authorities.

Pedro Oliveira of BusinessEurope also supported the business wallet proposal, describing it as one of the few EU digital initiatives likely to create tangible, practical value for businesses.

Oliveira pointed to fragmented company registration and beneficial ownership systems as examples of existing inefficiencies the wallet could help address. He also suggested future links between the wallet and due diligence obligations, Know Your Customer procedures, product passport systems, and possible EU company identifiers.

Several participants stressed that successful implementation will depend on integrating the wallets with existing infrastructure rather than creating parallel systems.

Jaromir Talir of the Czech registry CZ.NIC described how domain registries have struggled for years with reliable registrant identification, particularly in the context of cybersecurity and regulatory requirements. Talir said large-scale EUDI wallet pilots involving domain registries have already demonstrated use cases including registrant verification, authentication, and proof-of-domain-ownership credentials.

He also noted that some countries, including France and Denmark, already operate wallet-like systems while full certification processes continue.

A major part of the discussion focused on digital sovereignty and Europe’s dependence on foreign technology infrastructure.

Sebastiano Toffoletti of the European DIGITAL SME Alliance warned that wallet deployment still depends heavily on US-controlled mobile operating systems and hyperscale cloud providers. He argued that Europe should not build critical identity infrastructure on systems it does not control.

Toffoletti also suggested that large-scale wallet adoption could help support broader adoption of European digital alternatives by creating new distribution channels for European services.

Other speakers responded that the framework itself remains technologically neutral and does not mandate any specific infrastructure provider.

Meurens argued that the wallet framework could operate on different infrastructures if Europe chose to develop them, while Talir noted that alternative implementations, such as web-based wallets, remain possible.

The discussion also addressed concerns that the wallet ecosystem could ultimately strengthen the role of major technology companies if implementation becomes concentrated among dominant providers.

Vittorio Bertola of Open-Xchange warned that Europe could face two separate risks. Either the wallet fails to achieve adoption, or it succeeds but becomes dependent on large non-European firms capable of operating identity services at scale.

Sagstetter defended the wallet’s public-private model, arguing that successful adoption will require both trusted public oversight and useful private-sector services. He also stressed that Europe is contributing actively to the development of international digital identity standards rather than simply adopting frameworks developed elsewhere.

Several participants emphasised that European digital sovereignty should remain open and collaborative rather than protectionist or isolationist.

Oliveira described the concept as ‘open sovereignty,’ while Meurens argued that reducing vendor lock-in through interoperability, open standards, and diversification should remain the primary objective.

The session concluded with broad agreement that the EUDI wallet and the business wallet could become important infrastructure for the European single market if they achieve interoperability, security, trust, and practical usability.

Participants also agreed that adoption will depend heavily on whether the wallets provide convenient and widely used everyday services while preserving privacy, user control, and legal certainty.

EuroDIG 2026 took place on 26 and 27 May at the Charlemagne Building of the European Commission in Brussels under the theme ‘European Voices for the Future of the Internet – Celebrating 20 Years of .eu and the Beginning of a New Internet Governance Era’.

Digital Watch Observatory followed EuroDIG 2026 through a dedicated event page, featuring session information and reporting from Brussels.

Would you like to learn more about AI, tech and digital diplomacyIf so, ask our Diplo chatbot