commbox Artificial Intelligence & Algorithms

Artificial Intelligence & Algorithms

Filter

367

Sort by:

Direction:

External Link (174)

Preserving Human Voices and Faces. Message of his Holiness Pope Leo XIV for the 60th World Day of Social Communication

Leo XIV (Pope)

Vatican City: Holy See, January 24 (2026)

"The task laid before us is not to stop digital innovation, but rather to guide it and to be aware of its ambivalent nature. It is up to each of us to raise our voice in defense of human persons, so that we can truly assimilate these tools as allies. This alliance is possible, but needs to be based ... on three pillars: responsibility, cooperation and education. First of all, responsibility. Depending on the role we play, responsibility can be understood as honesty, transparency, courage, farsightedness, the duty of sharing knowledge or the right to be informed. As a general principle, however, no one can elude personal responsibility for the future we are building. For those at the helm of online platforms, this means ensuring that their business strategies are not guided solely by the criterion of profit maximization, but also by a forward-looking vision that considers the common good, just as each of them cares for the well-being of their own children. The creators and developers of AI models are invited to practice transparency and socially responsibility in regard to the design principles and moderation systems underlying their algorithms and the models they develop, in order to promote informed consent on the part of users. The same responsibility is also required of national legislators and supranational regulators, whose task it is to ensure respect for human dignity. Appropriate regulation can protect individuals from forming emotional attachments to chatbots and curb the spread of false, manipulative or misleading content, safeguarding the integrity of information as opposed to its deceptive simulation. [...] We are all called upon to cooperate. No sector can tackle the challenge of steering digital innovation and AI governance alone. Safeguards must therefore be put in place. All stakeholders — from the tech industry to legislators, from creative companies to academia, from artists to journalists and educators — must be involved in building and implementing informed and responsible digital citizenship. Education aims to do precisely this: To increase our personal ability to think critically; evaluate whether our sources are trustworthy and the possible interests behind selecting the information we have access to; to understand the psychological mechanisms involved; and to enable our families, communities and associations to develop practical criteria for a healthier and more responsible culture of communication. For this reason, it is increasingly urgent to introduce media, information and AI literacy into education systems at all levels, as already promoted by some civil institutions. As Catholics, we can and must contribute to this effort, so that individuals — especially young people — can acquire critical thinking skills and grow in freedom of spirit. This literacy should also be integrated into broader lifelong learning initiatives, reaching out to older adults and marginalized members of society, who often feel excluded and powerless in the face of rapid technological change. Media, information and AI literacy will help individuals avoid conforming to the anthropomorphizing tendencies of AI systems, and enable them to treat these systems as tools and always employ external validation of the sources provided by AI systems — which could be inaccurate or incorrect. Literacy will also allow for better privacy and data protection through increased awareness of security parameters and complaint options." (https://www.comunicazione.va/en/giornata-mondiale-comunicazioni-sociali/gmcs-2026/gmcs-2026-messaggio.html) more

External Link (213)

The Protocol Gap: Brazil

Sérgio Spagnuolo, Ester Borges, Bruno Fiaschetti, Paula Miraglia

Journalism Relay Project; Momentum; International Fund for Public Interest Media (IFPIM) (2026), 12 pp.

"An overwhelming majority of news websites in Brazil have either no AI-scraping policy at all or a very permissive approach. As of November 2025, only 7.2% of Brazilian news websites disallow at least one AI crawler through the use of a robots.txt file, even though most news sites (75%) have a robot ... more

Download (79)

Handbook on Data Protection and Privacy for Developers of Artificial Intelligence (AI) in India

Ajey Karthik, Aman Taneja, Sreenidhi Srinivasan, Vidushi Sinha

New Delhi: Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ); FAIR Forward – Artificial Intelligence for All (2026), 117 pp.

"This Handbook is designed as a practical guide for developers of AI systems, especially from early-stage startups, to navigate data protection obligations and ethical considerations, in a clear and actionable manner. Rather than serving as a comprehensive legal or technical manual, the Handbook co ... more

External Link (179)

Invisible Workers, Visible Harms: Perils and Precarities of AI Labour

Aapti Institute

Bonn: Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) (2026), 40 pp.

"Across South Asia, Southeast Asia, and Sub-Saharan Africa, data workers operate within fragmented and opaque supply chains characterised by informal employment arrangements, limited social protections, and a profound imbalance of power between workers, vendors, and the global technology firms that ... more

External Link (216)

Relational Futures: Indigenous Sovereignty and the Governance of Artificial Intelligence (AI)

Bronwyn Carlson, Tamika Worrell

Macquarie University (2026), 30 pp.

"Artificial intelligence must be governed as a relational and institutional force, not simply a technical system. Effective policy requires Indigenous leadership, enforceable standards, and a commitment to sovereignty, accountability, and community wellbeing. Failure to act risks embedding inequalit ... more

Download (117)

European AI Competitiveness Beyond Frontier Models: Insights From AI Startups in Five Strategic Markets

Nicole Lemke

Berlin: Interface – Tech Analysis and Policy Ideas for Europe (2026), 37 pp.

"AI development is frequently described as a race for frontier models. But considerable economic value will also be generated by AI development and adoption in existing markets. This data brief examines where EU competitiveness stands in the five applied markets where most of the EU’s AI startup a ... ctivity in our dataset takes place – horizontal enterprise software, healthcare and life sciences, manufacturing and supply chain, finance and insurance, and research and development tools. Drawing on data from Dealroom.co covering 24,468 active AI startups launched between 2012 and 2025 across the EU, Switzerland, the UK, and the U.S., we compare geographic patterns at the level of regions, countries, and cities through original analysis [...] Manufacturing and supply chain is the only market where the EU accounts for a larger share of AI startups than the U.S., though our data likely underestimates the number of U.S. startups in general. This suggests the competitiveness gap across applied AI markets may be even larger than our data indicates [...] AI startup activity in the EU is geographically dispersed: 75 percent of analysed EU AI startups are spread across 95 cities, with no single hub dominating any market. Paris, Berlin, Amsterdam, Munich, and Stockholm are the largest hubs but together account for only a quarter of EU AI startups. EU city rankings across markets largely reflect general AI startup hub sizes rather than genuine market specialisation – but specialised niches do begin to emerge at the local level, e.g. Munich in manufacturing and supply chain and Barcelona in healthcare and life sciences. As our data skews towards larger startup hubs, specialisation patterns in smaller cities and regions may not be fully captured. Our findings have important implications for policymakers. Europe's challenges in AI competitiveness extend well beyond frontier models. The manufacturing and supply chain market, a strategic priority for the European Commission, presents grounds for cautious optimism, where existing industrial strengths align with national and regional specialisation signals based on AI startup activity, especially in Germany and Munich. The geographically distributed startup activity in the EU can support AI adoption across diverse regional economies and existing industrial strengths. At the same time, it raises questions about whether European hubs can achieve the scale and specialisation needed to compete globally. The technological diversity of AI also presents competitive opportunities that go beyond market specialisation alone. Whether the emerging specialisation patterns identified in this analysis develop into globally competitive positions will depend on continued access to growth capital, compute resources, and cross-border market integration." (Executive summary, pages 3-4) more

Download (78)

Uso crítico y ético de la Inteligencia Artificial en la educación
Deep Insights

Santiago García Gago, Inés Binder

São Paulo: Campaña Latinoamericana por el Derecho a la Educación (CLADE); Educación en Voz Alta (2026), 150 pp.

"Esta guía sistematiza y resume los contenidos del curso virtual “Uso crítico y ético de la Inteligencia Artificial en la educación”, celebrado en tre los meses de octubre y diciembre de 2025 [...] El curso se planificó como una introducción conceptual a los distintos ámbitos que abarca ... la Inteligencia Artificial, entendida como un sistema sociotécnico complejo y multidimensional. La idea es que estos conceptos apoyen a docentes y educadores en el diseño de estrategias que incorporen la IA crítica y éticamente en los procesos educativos. Por lo tanto, no es manual técnico de aplicaciones ni tampoco uno de estrategias pedagógicas. Se concibió como un compendio de fundamentos teórico-prácticos que sustenten el diseño de las estrategias para incluir aplicaciones de IA en las aulas. Para ello, en el capítulo primero, se desmitifica el desarrollo tecnológico desde la sociología y la economía política preguntándonos quiénes son los dueños de la IA y qué intereses económicos y geopolíticos dirigen su destino. En el capítulo dos se abordan los conceptos técnicos, las definiciones y tipologías que permiten abrir la caja negra de la Inteligencia Artificial y entender cómo funciona. También se enumeran algunas aplicaciones, tanto las privativas como modelos open source que permiten experimentar en el ámbito educativo o usar con mayores niveles de privacidad estas herramientas. El capítulo tres, alejándose de la lógica instrumental que analiza las tecnologías desde los usos “buenos” y “malos”, problematiza las relaciones de poder que configuran la IA y los desafíos a los que nos enfrenta clasificados en siete dimensiones: técnicas, éticas, ecológicas, de género, políticas, laborales y de seguridad y privacidad. Y cierra con una breve exposición de los esfuerzos legislativos, tanto a nivel internacional como regional y nacional, para regular la Inteligencia Artificial. En el capítulo cuarto, se exponen diversos aspectos cognitivos y del El capítulo quinto, complementando el anterior, aborda estrategias concretas para la inclusión ética de la IA en los procesos de aprendizaje desde tres abordajes: con pedagogías centradas en la persona, integrales e interdisciplinares y basado en la tecnología y no en la aplicación." (Introducción, páginas 9-11) more

Download (128)

Societies at an AI Crossroads: Choices and Value Conflicts

Ignas Kalpokas, Julija Kalpokienė

London; New York: Routledge (2026), 116 pp.

"As contemporary societies undergo profound transformations driven by artificial intelligence and related technologies, a unilinear vision of inevitable ‘progress’ and development increasingly dominates public discourse. This book offers a counterpoint, challenging such deterministic narratives ... more

Download (23)

Taking a Look at the most common AI myths & debunking them the Superrr way!

Berlin: SUPERRR (2026), 22 pp.

"How do we want to use technology, and when do we hit the brakes? As part of our work in 2026, we held 14 public-facing “AI mythbusting” sessions at libraries across Berlin, Germany. We were astonished by the strong public response and clear need for judgement-free spaces to talk about AI. Colle ... more

Download (85)

Indic Voice Technologies for an Inclusive Digital India: Toolkit for Developers

Nihar Desai, Sujith P., Harleen Kaur, Dona Mathew, Jyotsna Jayaram, Padmavathi Prasad, Thomas Vallianeth

Bhashini; Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ); FAIR Forward – Artificial Intelligence for All (2026), 70 pp.

"This toolkit sets out a layered, lifecycle-oriented approach to building inclusive and robust speech artificial intelligence (AI) systems. It brings together strategies for diverse and representative data collection, linguistically informed model training, rigorous quality control, and deployment o ... more

External Link (223)

Toward AI Governance Alignment in Africa, Middle East, and Türkiye (AMET) Region

Emma Ruttkamp-Bloem, Fola Adeleke, Jake Effoduh, Kebene Wodajo, Nagla Rizk, Olubayo Adekanmbi, Rachel Adams, Rami Alkarmi, Seani Rananga, Seydina Ndiaye, Shikoh Gitau, Tami Koroye

Global Center on AI Governance (GCG) (2026), 54 pp.

"The analysis begins by identifying the core “centric approaches” that guide Artificial Intelligence (AI) national strategies. These include security-anchored models driven by state control, innovation-led frameworks that prioritise competitiveness, rights-based regimes centred on fundamental pr ... otections, and developmental approaches focused on public service delivery and technological catch-up. Each approach expresses incentives, risk perceptions and administrative capacities. Out of these foundational choices, there are different groupings. Various regions within AMET are adopting recognisable patterns of regulation that reflect their political economies, institutional structures and geopolitical positions. These groupings are not static categories, but evolving alignments shaped by domestic pressures, supply-chain dependencies and shifting global power dynamics. Across all regions, a set of cross-cutting issues consistently shapes the regulatory landscape. Data protection and interoperability remain central, alongside growing demands for accountability, explainability and robust auditing. Standards and certification are emerging as tools for both trustbuilding and geopolitical influence. Sector-specific rules in finance, health, education and public administration are increasingly decisive. Capacity constraints, fiscal limits and uneven institutional strength continue to define what is realistically possible for many governments. Geopolitical tensions and cloud concentration reinforce questions around sovereignty, infrastructure control and long-term strategic autonomy. This report translates these insights into practical implications for states, regional bodies and technology firms. It highlights workable pathways for reconciling innovation with safety, managing cross-border data flows, and building resilient digital infrastructure. It also identifies areas where architecture choices such as regional cloud centres, culturally calibrated testing layers and tiered sovereignty models can support both regulatory goals and developmental priorities. The recommendations emphasise flexible alignment, cooperation through standards, and the need for shared mechanisms that accommodate political diversity while reducing fragmentation. Taken together, the framework provides a stable and operational way of understanding global AI regulation. It clarifies the interests driving different jurisdictions, the pressures that shape convergence or divergence, and the strategic considerations that technology firms and policymakers must navigate as the regulatory environment continues to evolve." (Executive summary) more

Download (95)

Six Critical Lenses on AI-Generated Images

Catherine Bouko, Nataliia Laba (eds.)

CRC Press (2026), xvii, 176 pp.

"This book offers six critical, interdisciplinary approaches to AI-generated images. It invites the reader to (re-)imagine relationships between artefacts and media, authorship and agency, creativity and commodification, representation, and visual style. Across the chapters, the book examines three ... more

Download (104)

Environmental Cost of AI's Energy Use: Carbon, Water and Land Footprints

Miriam Aczel, Sanaz Chamanara, Mir Matin, Aria Farsi, Tshilidzi Marwala, Kaveh Madani

Richmond Hill, Ontario (CA): United Nations University Institute for Water, Environment and Health (UNU-INWEH) (2026), 55 pp.

"[...] The study reveals some striking numbers. In 2025, data centers—the physical backbone of AI—consumed an estimated 448 TWh of electricity. If data centers were a country, that level of electricity use would rank it 11th globally. On current trajectories, data center electricity demand could ... roughly double to 945 TWh by 2030, nearly triple the combined annual electricity use of Pakistan, Bangladesh, and Nigeria, together home to more than 650 million people. Producing that much electricity would have a carbon footprint of 399 million tonnes CO2e, requiring 6.7 billion trees grown over 10 years to offset—roughly twice the number of trees in the United Kingdom. The associated water footprint of 9.3 trillion liters would be equivalent to the annual domestic water needs of all 1.3 billion residents of Sub-Saharan Africa. The land footprint associated with this electricity would exceed 14,500 km², nearly 10 times the size of Mexico City. AI is now one of the most significant drivers of that data center growth. In 2025, AI workloads alone accounted for around 20% of total data center electricity use, and if that share rises to 40% by 2030 as projected, its electricity demand could reach roughly 378 TWh—enough to meet the residential electricity needs of the entire population of Sub-Saharan Africa for over 2 years. AI’s environmental impacts are shaped not only by data center growth and electricity supply mixes, but also by the escalating cost of building ever-larger models. For example, GPT-3 training consumed an estimated 1.3 GWh of electricity over 34 days, while GPT-4 is estimated to have consumed 50 to 70 GWh over 100 days, roughly 40 to 55 times GPT-3. Yet training is only part of the picture as AI’s operational footprint is increasingly driven by inference. Once models are deployed, billions of everyday interactions account for the bulk of energy use, with inference estimated at roughly 80–90% of total energy consumption. [...] The report identifies priority implications for key stakeholders. Governments must integrate AI infrastructure into energy system planning, carbon accounting, water governance and land use permitting. This requires standardized environmental footprint reporting so impacts can be verified and compared across providers and jurisdictions. Industry and AI developers should treat model selection, default outputs, and routing decisions as footprint determinants, while also improving efficiency-by-design. Users and deploying organizations also shape impacts through volume, frequency, and modality choices. They should adopt “fit-for-purpose” use: selecting the lightest model and lowestenergy format that meets the task and limiting high-cost features when not needed. Data center operators and utilities should recognize siting and procurement as environmental footprint decisions, apply environmental impact screening and cumulative impact assessment. They should implement transparent mitigation and community safeguards where expansion occurs. Investors and financiers should treat electricity, carbon, water, and land footprints as material risks for AI infrastructure portfolios and use comparable footprint metrics in due diligence." (Executive summary, pages 12-13) more

Download (100)

From Crisis to Renewal: Addressing AI’s Impact on Our Information Ecosystem

Lena-Maria Böswald, Roa Powell, Tyreese Calnan

Berlin: Interface – Tech Analysis and Policy Ideas for Europe (2026), 18 pp.

"As AI summaries summarise information directly within search results, they can significantly reduce traffic to news publishers. Evidence suggests users are less likely to click through to original sources when AIOs are present on Google, threatening the financial viability of journalism. These impa ... cts from AI summaries and AIOs in particular are unlikely to be evenly distributed and will play out in different ways across the news sector across Europe, with smaller and ad-funded outlets particularly vulnerable, raising concerns about declining media plurality. At the same time, AI-generated summaries often fail to provide accurate and diverse information. Research shows that AI tools disproportionately rely on a narrow set of sources, marginalising smaller publishers. In addition, persistent issues with accuracy – including misleading or incorrect outputs – risk distorting public understanding and weakening trust in information systems. Current regulatory frameworks in both the EU and UK are struggling to address these challenges effectively. While multiple legal instruments are relevant and already in place – including competition law, platform regulation, copyright, and AI governance – none are fully equipped to respond quickly and comprehensively to the harms of monopolisation and the lack of remuneration for media outlets posed by AI summaries. We argue that competition policy offers the most effective and immediate route to intervention, with digital regulation serving as a secondary lever. The UK’s approach, based on flexible, case-specific competition rules, enables faster action through tailored conduct requirements for dominant firms like Google. In contrast, the EU’s framework is broader but slower, relying more heavily on traditional, time-intensive antitrust investigations and ongoing debates around copyright and remuneration for publishers. Both systems offer complementary strengths, and policymakers should learn from each other’s approaches. We therefore recommend a “competition-first” policy package to ensure fairer outcomes. Key measures include enforcing fair ranking and attribution practices, preventing unauthorised use of publisher content, and potentially requiring platforms to compensate news organisations. However, competition policy alone is not the key. We therefore propose additional long-term interventions, including greater transparency through “nutrition labels” for AI-generated content and increased public or democratic funding for journalism. These measures aim to strengthen the resilience and independence of the news sector, ensuring that trustworthy information remains accessible in an information ecosystem shaped by AI. more

External Link (110)

We Have an Artificial Intelligence Definition Problem in ICT4D

Wayan Vota

ICTworks (2026)

"A new study in the International Journal of Research and Innovation in Social Science surveyed 200 women micro-entrepreneurs in Bengaluru and reported that 32.5% of them are “actively using at least one AI-enabled tool” in their businesses. The tools, according to the paper, include KhataBook ( ... more

Download (151)

Reinventing Humanitarian Aid Procurement for the Age of AI

Giulio Coppi

Access Now (2026), 59 pp.

"Our research reveals that the organisational use of AI and algorithmic systems often happens through a process of unstructured integration rather than formal adoption. More precisely, cloud-based, algorithmic-enhanced functionalities and processes are making their way into humanitarian entitiesʼ t ... ech stack through updates and add-ons to preexisting products, slowly but inexorably seeping into most internal systems. A price-based procurement process designed around the static idea of digital asset purchase is inadequate to ensure adequate protection from hyperdynamic tech product development. This is especially evident looking at the growing power imbalances between private actors and nonprofits, and among nonprofits themselves. Digital technology is often heralded as key to democratise and equalise the playing field between humanitarian actors, by finally unlocking local actorsʼ access to resources traditionally reserved for international entities. But emerging trends seem to disprove such claims. The lack of adequate legal frameworks and safeguards, lagging investments in humanitarian funding for information and communication technology (ICT) resources and connectivity, and inadequate dedicated funding for digital programmes are thwarting these opportunities. More broadly, virtuous and responsible procurement practices are proving inadequate for protecting human rights alone, as tech companiesʼ commitments to human rights are replaced by defence-oriented pitches and the size of their human rights teams keeps shrinking. Preexisting digital divides are enhanced by several deep and subtle changes in traditional tech business models, resulting from the gradual cloudification of most common digital systems and the fast-paced, distributed nature of modern digital development. Even more worryingly, we are seeing new forms of digital divide well beyond the traditional gaps in digital capacity between local and small non-governmental organisations (NGOs) and international non-governmental organisations (INGOs) or UN entities. Cloud-based companiesʼ takeover of generic coordination platforms creates a constantly broadening chasm between the few organisations that are digitally mature, those who are barely retrofitting their processes to fit the new reality, and the broad majority forced to access the promised land of digital humanitarianism through the predatory gates of algorithmic and cloud servicesʼ free tech licensing. Our findings highlight how organisation-led digital development, building on open-source systems (despite some “openwashing” practices) and trusted external providers, remains the best option for risk reduction when it comes to algorithmic systems. However, in the short to medium term, this might not be financially and technically feasible for the majority of humanitarian actors, especially considering the catastrophic funding environment. To fill this gap, we propose a toolkit in an appendix to this report. This is a foundational governance model for transforming digital procurement from a transactional into a strategic business function, integrated across key organisational departments. By doing so, protection-mandated aid groups could move from running purchase-centred processes, to adopting a dynamic, proactive, and rights-based approach to tech management. This report also contains a set of recommendations for states and donors, the humanitarian community, tech companies, local aid actors and communities, research centres, and cyber experts." (Executive summary, pages 2-3) more

Download (25)

World Development Report 2026: The Promise of Artificial Intelligence
Deep Insights

Washington, DC: World Bank (2026), xxxi, 341 pp.

"This Report is the first comprehensive assessment of what artificial intelligence (AI) means for 6.8 billion people living in low- and middle-income countries. It argues that AI represents a historic opportunity for these countries, potentially enabling them to overcome development challenges that ... have defied solution for decades. Given the scarcity of information on AI’s use in developing countries, the Report sets the ball rolling on closing the gap. It uncovers new facts about how businesses and governments in developing countries are using AI—and identifies the main obstacles to greater and more productive use. AI is a general-purpose technology, or GPT. That puts it in rare company. Steam power. Electricity. Computers and the internet. Each transformed not just one sector, but entire economies, and did so continuously, over decades. Each also followed a pattern: slow and uncertain at first, but eventually reshaping how economies produce and how societies function. AI is following a similar pattern, but on a much faster timeline. It took about 80 years for the steam engine to reach lower-income countries, 40 years for electricity, and 20 years for the internet.1 By contrast, middle-income countries accounted for half of ChatGPT’s global traffic within six months of its launch.2 Rapid diffusion on such a scale has not been seen before. But history also teaches us that the countries that benefited most from previous GPTs were not always the ones that came up with the underlying technology. Those that benefited had built the necessary foundations and moved fast to develop applications on top of them. The countries that did not invest in the infrastructure, skills, and institutions that such technologies required to function benefited less and felt the most pain from disruption. This lesson is directly relevant to AI: The winning approach will take advantage of the new technologies, not seek refuge from them. In only a few years, AI has evolved from making predictions to generating content from text, audio, and video to now autonomously executing complex tasks. As it becomes more powerful and pervasive, its value as a tool for development depends on three factors. The first is capabilities. Unlike computers or the internet, AI can perform complex thinking tasks that guide decision-making and usually require human expertise, such as diagnosing diseases, forecasting weather, or advising farmers. These skills are often in short supply in developing countries, so AI has the potential to fill important gaps. The second is concentration. The most advanced AI models, the chips they rely on, and the data centers that run them are controlled by a small number of companies in just a few economies. This concentration creates dependency risks and affects whether the AI tools available will meet developing countries’ needs. However, this concentration also allows countries to customize AI models without having to spend billions of dollars creating the most advanced AI systems from scratch. The third is complements: the foundations that allow AI to be used safely and effectively. AI works best where reliable infrastructure, good education systems, and strong institutions are present. Developing countries often have shortages in all three. AI, moreover, needs to be customized to the local context because it learns from the data upon which it is trained. How quickly countries build these complements—and how well governments and businesses reorganize to use AI productively—will determine how much and how fast AI improves development. In that respect, AI is no different from GPTs that have come before. Ten years ago, World Development Report 2016: Digital Dividends called these prerequisites the analog complements to digital technologies. Today, AI requires both strong analog and digital complements to deliver its full benefits." (Overview, pages 1-2) more

External Link (206)

Building an Open and Responsible Voice Technology Ecosystem: Policy Recommendations for Digital Inclusion in India

Bhashini; Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ); FAIR Forward – Artificial Intelligence for All (2026), 57 pp.

"This report examines key barriers to building open and responsible speech systems in India—from data collection and model development to infrastructure and responsible practices. It proposes policy recommendations and governance mechanisms to support an innovative and equitable voice-technology e ... cosystem. Key Recommendations: A. TREATING FOUNDATIONAL DATASETS AS PUBLIC GOODS: Foundational datasets for speech technologies are large, reusable corpora of audio, text, and metadata. They are curated to support a wide array of downstream applications, including automatic speech recognition (ASR), text-to-speech (TTS), and speech translation. Making foundational datasets available as digital public goods addresses market failures in voice-technology ecosystems and promotes local economic innovation. The report identifies challenges arising from the linguistic diversity and nuance of Indian languages, infrastructural barriers, the absence of common data standards that create governance gaps, and unresolved intellectual property and privacy issues. Policy recommendations for building foundational language datasets include clarifying and revisiting existing laws to enable the use of publicly available material, ensuring sustainable investments supported by government and blended finance techniques, and instituting strong governance systems with shared standards, coordinated repositories, and independent quality assurance. B. BUILDING OPEN AND REPRESENTATIVE MODELS: India’s success in inclusive voice technologies depends on whether speech systems perform well across the country’s linguistic and social diversity. Today, limited openly available datasets, inadequate evaluation benchmarks, and uneven access to compute have resulted in models that perform inconsistently across languages, accents, and demographic groups, especially those from rural, low-resource, or marginalised communities. Strengthening India’s ecosystem requires further investment in data, evaluation, and infrastructure. Benchmarking should be built around open evaluation datasets and transparent leaderboards providing common baselines for developers, improving procurement standards, and helping assess performance across diverse languages and speaker groups. On the infrastructure side, platforms such as Bhashini, ULCA, and AI Kosh offer strong foundations although effective operation hinges on sustained governance, clear access protocols, and long-term funding models. C. INSTITUTIONALISING SUSTAINABLE OPEN-SOURCE INFRASTRUCTURE: Speech datasets place far greater demands on storage, bandwidth, and compute than text data, making the financing and governance of longterm hosting a central challenge for India’s voice-tech ecosystem. A fragmented licensing landscape adds further uncertainty: overlapping or incompatible terms across data, code, model weights, and evaluation sets impose substantial compliance burdens on small actors, while enforcement gaps allow misuse with little recourse. Sustaining open, equitable development requires treating dataset hosting as durable public digital infrastructure rather than grant-based, project-specific assets. By international standards, India’s emerging platforms, such as AI Kosh, provide a remarkable foundation. However, they require longterm funding, transparent governance, and clear access pathways for non-government contributors. Collaborative stewardship models, such as the Mozilla Data Collective, can help establish shared quality norms and consistent metadata conventions. D. STRENGTHENING RESPONSIBLE DEPLOYMENT: Deploying speech technologies responsibly requires more than high-performance models; it depends on safe systems, contextually appropriate use, and clear accountability. Existing data practices lack value-sharing mechanisms, leaving communities and researchers without recognition or benefitsharing, even as their contributions fuel commercial products. Risks of misuse, including voice cloning, phishing, and deepfake-driven misinformation, are rising, and unintended harms like linguistic exclusion, biased performance across accents and genders, and the erosion of regional language identities remain widespread. Addressing these gaps requires structural guardrails that embed fairness, transparency, and accountability into deployment workflows. Preventing misuse demands a combination of technical safeguards, stronger legal pathways, and widespread public literacy efforts to help users recognise risks and exercise their rights. The report argues that a strong ecosystem requires more than innovation funding. Building open-source foundations—including language datasets, standards, collection protocols and responsible AI frameworks—promotes demand-driven local innovation. It is therefore essential that the state plays an active, shaping role, much as it has in the development of digital public infrastructure. In the context of voice technology, this involves both investing in commercially viable languages and sustaining low-resource languages that are vital for inclusion but unlikely to attract private capital. Open-source assets can reduce costs for the public and private sector alike. However, they demand long-term planning and financing for hosting, maintenance, and updates. These assets can be supported through blended-finance models that pool public, philanthropic, and commercial resources. Emerging national initiatives, such as the proposed AI marketplaces, can further structure participation, transparency, and value-sharing across data, annotation, and deployable models." (Executive summary, pages 9-11) more

External Link (222)

AI is here. Are we ready?

Eva-Maria Verfürth et al.

D+C: Development and Cooperation, issue 4 (2026), pp. 31-58

"AI is already having a profound impact on our daily lives. The key question is who controls the technology. It is not only in conflicts that AI accelerates the spread of deceptively realistic fakes and makes it harder to distinguish reliable information from manipulation. At the same time, it opens ... more

Download (98)

Companion AI: Risiken von sykophantischen und suchtfördernden Designs von KI-Systemen, ihre rechtliche Erfassung und Handlungsempfehlungen
Deep Insights

Ramak Molavi-Vasse’i

Berlin: Zentrum für Digitalrechte und Demokratie (2026), 82 pp.

"Diese Studie untersucht Risiken von KI-Systemen, die im fortlaufenden Dialog mit Nutzern Emotionen erkennen, persönliche Bedürfnisse adressieren und darauf reagieren. Chatbots mit dieser Funktionalität werden als Companion-AI bezeichnet. Sie sind hochgradig personalisiert und treten Nutzern als ... soziales Gegenüber entgegen, das freundschaftliche, romantische oder sexuelle Nähe simuliert. Menschen können zu solchen Systemen eine emotionale Bindung aufbauen. Der Fokus liegt nicht nur auf spezifischen Companion-Anwendungen wie Replika oder Character.AI, sondern auch auf Universalmodellen wie ChatGPT, Claude, Gemini, Grok oder Meta-AI. Diese Systeme werden zunehmend für persönliche, emotionale und beratende Gespräche genutzt. Die Untersuchung kommt zu dem Ergebnis, dass Companion-AI mehrere eng miteinander verbundene Risikodimensionen erzeugt. Psychische und physische Gesundheit: Companion-AI trifft auf eine Gesellschaft, in der Einsamkeit und psychische Belastungen zunehmen, während professionelle Versorgungsangebote knapp sind. Die möglichen Risiken sind klinisch belegt. Die Nutzung kann psychische Belastungen verursachen oder verstärken und in Einzelfällen gravierende Gesundheitsfolgen auslösen. Dokumentiert sind die Verschlimmerung psychotischer Zustände, die Verstärkung depressiver Muster und Angststörungen, suchtartige Bindungen mit Entzugssymptomen sowie die Erosion sozialer Kompetenzen, etwa eine messbar reduzierte Konfliktfähigkeit nach längerer Interaktion mit Companion-AI. Öffentlich bekannt gewordene Vorfälle sind in der Vorfall-Datenbank der Studie dokumentiert. Eingriff in die Privatsphäre: Companion-AI animieren Nutzer kontinuierlich zur Selbstoffenbarung und greifen damit in sensible Gedanken und die intime Gefühlssphäre der Nutzer ein. Zugleich ermöglicht die fortlaufende Interaktion eine zunehmend verdichtete Profilbildung. Entscheidungsautonomie und demokratische Meinungsbildung: Companion-AI können sowohl die Qualität von Informationen als auch die Entscheidungsautonomie der Bürger beeinträchtigen. Die Mechanismen, die Nähe und Vertrauen erzeugen, beeinflussen zugleich die Generierung, Aufnahme und Gewichtung von Informationen. Sykophanz, also die unkritische Bestätigung von Nutzeransichten, beeinträchtigt Genauigkeit und Verlässlichkeit der Antworten messbar. Sprachmodelle werden zunehmend als primäre Quelle der Informationssuche genutzt. Wenn dieselben Systeme Informationen erzeugen, beschaffen und präsentieren, fallen Selektion und Aufbereitung in einer Hand zusammen. Werbe- und interessengeleitete Einflussnahme greift dann nicht mehr nur in einzelne Kaufentscheidungen ein, sondern in die Voraussetzungen öffentlicher Meinungs- und demokratischer Willensbildung." (Executive summary, Seite 5) more

Uso crítico y ético de la Inteligencia Artificial en la educación Deep Insights

World Development Report 2026: The Promise of Artificial Intelligence Deep Insights

Companion AI: Risiken von sykophantischen und suchtfördernden Designs von KI-Systemen, ihre rechtliche Erfassung und Handlungsempfehlungen Deep Insights

Uso crítico y ético de la Inteligencia Artificial en la educación
Deep Insights

World Development Report 2026: The Promise of Artificial Intelligence
Deep Insights

Companion AI: Risiken von sykophantischen und suchtfördernden Designs von KI-Systemen, ihre rechtliche Erfassung und Handlungsempfehlungen
Deep Insights