Anglia Ruskin University and Hughes Hall, University of Cambridge

The first Global Dialogue on AI Governance presents a key opportunity, to reflect and redress the ultrarapid diffusion process of AI technologies. The only meaningful approach on AI governance is naturally transnational, since a few AI large technologies firms are creating and transforming modalities of work, industries, information, markets across multiple sectors and countries. A global reflection by a wider set of stakeholders, focussing on the processes, impact and implications of AI diffusion is a necessity, as the current market incentives, while very successful in starting up the innovation processes are leading to increased market power, and multiple market failures and or environmental and social externalities. A diffusion process characterised by increasing returns, initially due to access to valuable personal data, used in the training of the algorithms is likely to lead to market tipping, negative social externalities, and increased inequalities across all sectors of life, at global scale. A success would be in the identification of the key risks, due to the specificity of AI economics and to have an international coordination forum to set framework necessary to protect, societies and economies from the possible outcomes of unbridled market power.

• AI capacity-building
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Safe, secure and trustworthy AI

Safe, secure and trustworthy AI is a precondition required for even starting reasoning within a global dialogue on AI. This is self-evident, given the power of teh instrument, an unsafe and unsecure and untrustworthy AI, risks become a nightmare for humanity, and the risk is not negligible. Ai Capacity building, is essential at global scale, not to leave regions areas left behind, especially for teh reallocation of labour demand across sector an countries. The Social, economic, ethical, cultural, linguistic and technical implications of AI, are, once the first two themes are addressed teh key focus of the dialogue, as these are the implications that a pure market value driven approach would often neglect, with the risk of resulting in a net negative impact on humanity. Transparency, accountability, and human oversight, are all key conditions to understand the working algorithms, and for example to discover possible perverse design aiming at limiting competition, abusing market power, create addiction. No regulatory oversight is possible without such transparency.

The theme of Market power is not well captured and i believe, that it affects and is affected by all the above themes. To achieve Market power key players might be tempted to bypass "Safe, secure and trustworthy AI", to retain in house "AI capacity-building" only, they will not internalise any incentives to address non profit related "Social, economic, ethical, cultural, linguistic and technical implications of AI". The would exploit "Interoperability of governance approaches" to extend market power across different jurisdictions. Market power is indifferent, and often detrimental to the "protection and promotion of human rights". Market power tends to abhor and contrast any effort to "Transparency, accountability, and human oversight" and identifies "Open-source software, open data and open AI models" as a risk that could erode market power.

I will now focus on a narrow issue, where my contribution focussed, while of course, i retain the belief of the the wider impact as discussed above. I will base the next reflections on a policy brief "Energy Data Spaces and Market Power: a new challenge for data sovereignty and its governance" (Giovannetti 2026) published by LSE [available at https://www.lse.ac.uk/ideas/publications/policy-briefs/energy-data-spaces-and-market-power-a-new-challenge-for-data-sovereignty-and-its-governance] that examines how the digitalisation of energy markets, driven by smart meter data and Energy Data Spaces, is reshaping competition in retail electricity services. It analyses the implications for market power, consumer outcomes, and data sovereignty, highlighting the regulatory challenges posed by data-driven business models that operate across national boundaries. In particular, the integration of digitalisation and artificial intelligence into energy retail markets, brings out additional fundamental questions, as AI enables real-time optimisation of retailers' tariffs based on predictors – such as fine-grained consumption patterns, weather forecasts, and market signals. These models can be designed to allow for different dynamic pricing, segmenting consumers based on behavioural flexibility and demand elasticity, and offering personalised pricing that reflects individual usage profiles. While this can enhance efficiency and consumers' satisfaction, it also introduces new risks, including opacity in pricing logic, potential bias against less flexible consumers, and market concentration due to data driven advantages. These models can be used by energy retailers to develop different business strategies, such as tailoring, bundling and dynamic pricing with the aim of locking-in existing customers with their existing providers of energy services increasing their switching costs. This, in turn, leads to reduced inter-retailer and market price elasticity, even while facilitating intra-day and intra-retailer price elasticity. Business strategies, based on AI algorithms accessing large quantities of granular data, present two types of benefits: individual and societal ones. The individual benefits are based on the potential of dramatically improving the experienced quality of the services, and on providing means to optimise usage and costs of energy, while staying with an existing provider. The societal benefits are, instead, linked to the ability of incentivising optimal off/on-peak usage at very short time intervals, or intra-day demand flexibility within a contract; this facilitates the integration in the electricity grids of highly time-variable Renewable Energy Sources. As a result of these processes, granular data that are fed into algorithms to create 'intelligent' business strategies for energy retailers are particularly welcome. However, the duty of this dialogue is to go beyond the obvious first-degree benefits and to focus on the emerging trade-offs and possible unintended consequences, especially from a dynamic efficiency point of view. The business strategies, enabled by the digitalisation of Energy Data Spaces increase brand loyalty, hence customers satisfaction and willingness to pay that, in absence of effective competition, creates a surplus that can be entirely appropriated by the current retailer.

The AI Dialogue should focus on nurturing a truly competitive environment, protect entry and digital sovereignty. This is essential as Energy Digital Spaces enable business strategies which exploit the power of personal data to generate increasing return; the better and the more data are fed into an intelligent tariff algorithm used to provide tailored and bundled energy services, the more likely it is that a customer will stay with their original provider, further generating personal data, and therefore further increasing the retailer grip on their custom. The scope of sovereignty should not be based on national boundaries, due to the incredible relevance of scale economies, but should be extended to jurisdictions sharing agreed principles on international data governance. It should cover not only consumer protection but also definitions and restrictions on market dominance, by sharing and agreeing the criteria for identifying the emergence of gatekeepers and their obligations toward fair access to their data infrastructure by entrants and non-discriminatory access pricing, along with non-excessive consumers' pricing. The focus of the dialogue should also include the distinction between personal and derived data, ensuring that interoperability and associated data portability are technically feasible at low economic and cognitive costs. This is required since when personal data are merged across consumers, an algorithm can create even more useful intelligent outputs; the more customers, the more useful the data are for each single user. This process is a typically self-reinforcing process generating positive network externalities – further reducing the potential for innovators and entrants to compete in these markets. Focussed impact analysis on households should be performed to assess the social impact of the emergence of Energy data spaces, due to their differentiated impact on pricing and services. The quality benefits of improved services can be appropriate at different levels, by different categories of consumers – depending on their technology readiness levels, in terms of skills, time, income and availability of smart home features. Often, for example, smarter white good are less affordable than the less smart ones, whereby being smarter may be related to remote activation, required to make the most of intra-day flexible tariffs. Asymmetric consumers' levels of awareness, as well as relative affordability, might pre-empt effective competition. Hence, it is also essential to empower consumers by launching public education campaigns and digital literacy programmes, helping consumers understand smart tariffs and data usage, shifting the costs of smart tools adoptions from the consumers to the sector providers. Strategic impact analysis of the market impact of the emergence of Energy data spaces on market entry is also needed. Typical policy responses designed to deal with the risks of increasing returns are grounded in the promotion of Data Interoperability, mandating open standards for energy data exchange and ensuring fair access for all market participants. This clearly aims at reducing switching costs – as users might be porting their data – and possibly also do multihoming at reduced costs, without the need to duplicate fixed data costs; multiple competing providers might use their data, independently on whom they are customers of. However, Giovannetti and Siciliani (2020; 2023), showed that essential data portability might not be enough; when incumbents face the incentives to price more aggressively, in order to keep entrants away, this prevents their use of legally portable data.

This research stems from the work by the Horizon Europe project European Distributed Data Infrastructure for Energy (EDDIE), aimed at designing a decentralised, distributed, interoperable, and secure data infrastructure tailored to the energy sector. Its goal being to enable a secure exchange, accessibility, and sovereignty of energy data among EU actors. By decentralising control and ensuring compliance with EU regulations, project EDDIE focussed on technological and data autonomy, advancing digital sovereignty by enabling Europe to host and govern its own data infrastructure for energy, addressing the problems of naturally emerging alliances that may naturally relying on foreign cloud services. The adoption of a data architecture based on the key principles of federated identity, data interoperability, and governance frameworks controlled by European actors is essential to this purpose. Policies should therefore refine their ways to assess the impact of data-driven bundling and profiling on market concentration through a network approach (ITU-T, 2025). A network approach, focusing on all the interrelated digital value chain components that includes all services bundled into energy services, will also contain a key international dimension, requiring further analysis and possibly transnational regulatory harmonisation. Some of the bundled energy components are clearly non-local. Smart meter data and digitalisation offer transformative potential for the energy sector. However, without thoughtful transnational governance and harmonised regulation, these innovations may reinforce existing inequalities and market dominance. Policy and regulators must act decisively to ensure that digitalisation – while facilitating the green transition and integrating RES – also keeps serving the public interest and supports a resilient, consumer-centric energy system. Lastly, the EU experience offers an opportunity to reflect more globally upon how to address these challenges of inequality and market dominance. Some key references are: -Davi-Arderius, D. et al. (2025) 'Digitalisation and Beyond: Economic Perspectives on Granular Energy Data', ITU Journal of Future and Evolving technologies, 6-4, pp. 400-415. Available at: https://www.itu.int/pub/S-JNL-VOL6.ISSUE4-2025-A31. - Giovannetti, E. and Siciliani, P. (2023) 'Platform competition and incumbency advantage under heterogeneous lock-in effects' Information Economics and Policy, 63, p.101031. Available at: https://doi.org/10.1016/j.infoecopol.2023.101031. - International Telecommunication Union Report, (2025) "Economic aspects of national Telecommunications/ICTs" . ISBN 978-92-61-41041-4 (Electronic version). - Giovannetti, E., (2026) Energy Data Spaces and Market Power: a new challenge for data sovereignty and its governance, LSE Ideas Policy Briefs, January 2026. -ITU-T (2025) Focus Group on cost models for affordable data services. FG-CD-D6 Technical report on "Terminology and taxonomy for international internet connectivity".

Most citizens would have little awarness of the degree by which Ai driven algorithms forms their choices, both consumptions, labour market and political/opining ones. Hence the AI Dialogue should reach all, disseminate insight and collect opinions at large scale.

Users, digitally illiterate communities, discounted communities. They should be the focus of a major outreach process,

Large events are less suited for this reach, distributed community based engagement formats could more effectively increase participation to this existential Dialogue.

The questions suggested in the recommendations from Chapter 3 of the the ITU_D Q4.1 Report,(2025) on the "Economic aspects of national Telecommunications/ICTs" ISBN 978-92-61-41041-4 are a starting point to collect relevant policy information. Annex 1: Proposed additional questions to the ITU Tariff Policies Survey Section 1: Data portability , interoperability, and open access to data held by Gatekeepers Part A: Data portability policies 1. Does your country have specific regulations mandating data portability* across digital platforms? o Yes o No 1.1 If yes, which sectors are covered: € Finance (e.g.: Banking) € Telecommunications € Mail and communications platforms (e.g.: e-mail, messaging, chat) € Energy € E-commerce € Others, please specify: ______________ 1.2 If no, what barriers prevent the adoption of data portability laws? a. Technical barriers: € Lack of data standardization : There's no universal format for data portability. Companies store and structure data differently, making transfers across platforms complex € Legacy systems: Older IT infrastructures may not support easy data export or interoperability € Data complexity: User data can include metadata, behavioural data, or inferred preferences, which may not translate easily between platforms b. Regulatory and legal challenges: € Unclear scope: it is ambiguous what data should be portable (e.g.: raw data, insights, recommendations) € Intellectual property concerns (e.g.: some data may mix personal information with proprietary algorithms or third-party content) 2. Does your country have mechanisms to ensure data transfer between service providers while maintaining user control and privacy? o Yes o No 2.1 If yes, are these: € Interoperable data standards (e.g.: common formats and application programming interface (APIs) to ensure data can move smoothly across platforms) € Strong user authentication (e.g.: multi-factor authentication to confirm the identity of the data requester before any transfer) € Consent management frameworks € Encryption in transit (e.g.: using transport layer security (TLS)) € Others, please specify: ______________ 3. Are there any financial or technical support systems in place in your country to help smaller businesses comply with data portability requirements? o Yes o No 4. Does your country enforce portability of all relevant personal data, including inferred data (e.g.: algorithm-based recommendations, personalized settings)? o Yes o No Part B: Data interoperability 5. Are there national or sector-specific standards for interoperability of digital platforms and data exchanges in your country? o Yes o No 6. Does your country have a regulatory and/or policy framework for implementing the governance of secure and privacy-preserving data interoperability? o Yes o No 6.1 If yes, is compliance monitored and enforced by a dedicated authority? o Yes o No 6.2 If yes, which authority? ______________________ Part C: Open access to data held by Gatekeepers 7. Does your country have regulations that require large digital platforms (gatekeepers) to allow portability of data (users can move their data to another service)? o Yes o No 7.1 If yes, how are these regulations structured: o Mandatory access o Voluntary compliance (e.g.: bilateral agreements) o Licensing models 7.2 If no, are there ongoing discussions or initiatives to introduce such measures? o Yes o No 8. Does your country have regulations to prevent dominant platforms from using exclusive data access as a competitive advantage over smaller businesses and star-tups? o Yes o No 9. Are there any mandatory data-sharing requirements for large digital platforms or gatekeepers in sectors critical for public interest (e.g.: health, finance, energy, transportation)? o Yes o No 10. Does your country have a mechanism for resolving disputes related to data access between gatekeepers and third-party service providers? o Yes o No ______________________