Original publication here.
The irruption of Generative Artificial Intelligence (GenAI) represents a watershed moment for public interest media and democracy. As we argued in a previous article, the extractive and concentration-based practices of big Generative AI providers, coupled with inadequate regulation and a media ecosystem already weakened by existing trends towards platformisation, have implications for the sustainability and integrity of public interest media and knowledge ecosystems, key pillars of our democracies.
As part of the ‘Strengthening Media in the Age of Generative AI’ action-research and dialogue, The Data Tank conducted a literature review which included a total of 221 academic, industry, and policy sources (you can read more details about the methodology in the report). The aim was to map the challenges that Generative AI poses to public interest media in Europe –and to small and medium-sized organisations in particular– and the ways forward being proposed to address them.
The challenges and the potential solutions are multiple and interconnected. Based on our thematic analysis of the literature, we present them as diagnosis of the current context on the one hand, and as potential ways forward on the other, acknowledging, however, that there are no such neat or binary distinctions: the limitations of some proposed solutions can amplify the challenges identified. For example, the review finds that bilateral licensing deals between large media publishers and large AI corporations are not seen to help small and independent media and do not solve the issues with misattribution and unreliability of Generative AI outputs (BBC and EBU, 2025).
Foster media sustainability and information integrity
As we write in the report, the challenges driven by Generative AI are multiple but, importantly, ‘are seen to affect the whole political economy of the media and information ecosystem, in a way that cannot be fixed by technical solutions, literacy and skills, or bilateral licensing deals alone’.
The impact of AI crawlers and content extraction without compensation is taking place at scale while audiences are also driven away from the media where their content is sourced to AI-summaries and multimodal tabs in AI assistants. This scraping is also placing a strain on the technical infrastructure of media outlets, which smaller organisations are less resourced to handle. This is impacting on the sustainability and the business models of media organisations.
At the same time, the quality of public interest content and the knowledge ecosystem is threatened as Generative AI assistants are inherently error-prone as they operate on probability-based inference rather than cognitive capabilities, unable to reliably provide the nuance, timing, context, attribution and verification that are essential to journalistic content and public interest media. The editorial gatekeeping processes and oversight mechanisms that exist for public interest media have its flaws, but provide layers of assurance and mechanisms for transparency and redress. By contrast, AI systems bypass the technical and governance standards and institutions that mediate the public sphere or, with perhaps less success, the increasingly complex or insular networked sphere. While this process of disintermediation had already worsened with social media networks and large platforms, AI systems are amplifying and complicating already concerning trends and their impact on democracy.
Evidence so far suggests that Generative AI is more likely to use English-language text, even in contexts where English is not the first language (Brantner et al., 2025), which, coupled with the sustainability issues, risks causing ‘news deserts’ (Mansell et al., 2025). Ultimately, as we write in the report: ‘The growing concentration of Generative AI power in a few hands, as well as concentration in ways information is generated (with English-centric and large national media favoured over more local or diverse sources), further erodes pluralism, equity, and diversity’.
Approaches to regain bargaining power
A central point in the literature and the diagnosis of the situation refers to what is considered to be a loss of leveraging power and an inadequate regulatory framework, not fit to deal with these structural impacts. On the loss of bargaining power, direct bilateral deals are out of reach for smaller media outlets, impact brand and trust and do not address issues of information integrity and pluralism. Yet, collective licensing such as the case of the Danish Press Publications Collective Management Organisation exemplifies, media collaboratives or approaches to statutory licensing point towards potential ways forward to reclaim agency and negotiating power. On regulation, proposals range from copyright reform to competition regulation, and include backstop arbitration or the establishment of levy taxes applied to Generative AI corporations, the revenue of which would be redirected back to public interest media actors, as proposed by Open Future (Keller, 2025) or as recently argued by Mistral (Mensch, 2026).
Collective action by media coalitions and public and philanthropic investment are viewed as essential. And while media, data and AI literacy for audiences and media professionals alike are necessary, they will not be sufficient on its own.
New visions and a holistic approach for the way forward
Our review of the literature sheds light on the complex interplay between Generative AI, the media ecosystem, the integrity of information and, ultimately, democracy. In trying to capture this complexity, a few overarching themes stand out.
The current changes driven by Generative AI technologies both reflect and result in power asymmetries rewiring how information is produced, interpreted, distributed and used, bypassing democratic mechanisms for shared knowledge, contestation and accountability. Beyond technical fixes and media literacy, there is the need for solutions to strengthen collective bargaining power, supported by regulatory initiatives and investment in public digital infrastructure. These holistic solutions require collective action and collaborative, inclusive ecosystems with resources that include and support regional, local, niche outlets.
Adaptive, modular, and scalable business models and strategies are needed, including diversification and approaches that create spaces of belonging and trust to keep audiences engaged. Yet, not enough is known at this point about what strategies will work. Successful approaches to economic sustainability require time, funding and visibility.
Public interest journalism and media remain essential to preserve information integrity, accountability, and the diverse tapestry of voices and experiences within our democracies. Democratic governance of media, AI platforms and technology providers would better align the existing tensions between public interest and the reduced agency that comes with AI system architectures.
A collaborative approach and a multiplicity of actions, from the micro to the macro level, including institutional imagination, will be the essential compass in this critical juncture.
