UNESCO’s A.I. Report Considers How Technology Can Serve Culture

UNESCO’s mission is not only to preserve the heritage left behind by previous generations, but also to identify, track and promote the circulation of knowledge and ideas that foster further cultural creation, development and dialogue. It’s a mission that inevitably straddles past, present and future, and so the organization could not avoid confronting one of the biggest elephants in the room today: the role of A.I. in reshaping how ideas circulate, develop and take form in contemporary culture. The latest UNESCO survey on A.I., prepared with the Artificial Intelligence and Culture (CULTAI) Independent Expert Group, gathers views and advice on the evolving relationship between culture and artificial intelligence.

The key finding is that A.I. is already transforming how culture is created, shared and preserved by expanding access, safeguarding heritage, amplifying diverse voices and opening new creative possibilities. At the same time, however, this technology can also deepen inequalities, intensify biases, accelerate cultural homogenization and increase environmental costs. UNESCO also acknowledges complex ethical and practical risks posed by A.I., which the organization had already begun to confront via its 2017 Operational Guidelines for updating the 2005 Convention on the Protection and Promotion of the Diversity of Cultural Expressions for the digital age and the landmark 2018 report, Re|Shaping Cultural Policies. Nearly a decade ago, those documents identified A.I. as a transformative agent in content generation, reframing the creative value chain and positioning data at the core of cultural policy. At the time, UNESCO asserted that the coming transformation would require a governance framework anchored in three essential pillars: transparency, participatory design and accountability.

In 2021, 194 UNESCO member states adopted the Recommendation on the Ethics of Artificial Intelligence—the first and, so far, only global normative instrument on A.I. that explicitly includes culture as a specific domain. The text provides essential ethical principles and advocates for creators’ rights in the A.I. context. In 2023, UNESCO further expanded this framework with guidelines for regulating digital platforms, urging safeguards for freedom of expression, access to information and culturally sensitive A.I. moderation, including bias audits across languages and contexts.

The latest report was part of the preparatory work for MONDIACULT 2025, held in Barcelona, which brought together member states to consider future global cultural policies.

The challenges of A.I., as outlined by UNESCO

The acceleration of A.I. and artificial intelligence-powered technologies is widely acknowledged to have already exceeded the capacity of cultural ecosystems to adapt, outpacing governance policies and potentially threatening sovereignty, pluralism and democratic oversight, while raising ethical questions about authorship and rights. At the core of UNESCO’s report is an urgent call for global cooperation and inclusive governance. While some regional or local initiatives and policies are already being implemented, the global nature of these transformations makes it essential that all countries develop national A.I. strategies that incorporate the cultural dimension and address concerns like:

The reinforcement of bias and homogenization

Unbalanced training data and algorithmic personalization reproduce stereotypes and foster monocultures, obscuring cultural distinctiveness, widening inequalities and even exacerbating injustices. The document highlights how unequal control over A.I. tools and platforms, once combined with disparities in digital technologies, skills and knowledge, can undermine the very principle of culture as a global public good. As A.I. systems are largely trained on “collective commons,” they tend to perpetuate biases and stereotypes rooted in dominant epistemologies. Several data points in the survey bear this out: Midjourney, for example, represents women in only 23 percent of images and people of African descent in 9 percent, compared with U.S. labor market data, in which women account for 46.8 percent and Black people for 12.6 percent. A.I.-generated content tends, for now, to amplify or exacerbate existing disparities and inequalities, particularly when it comes to systems trained on synthetic rather than human-generated content.

Widening disparities and undermined cultural rights

Limited technical training can make it harder for parts of the cultural sector to adapt to an A.I.-driven environment, further exacerbating existing inequalities. According to a 2024 CISAC report, music sector workers could lose nearly 25 percent of their income to generative A.I. by 2028, while those in the audiovisual sector may face a 20 percent drop. Meanwhile, the exploitation of unprotected cultural data without consent, attribution or compensation, along with the marginalization of minority cultures, raises urgent concerns. Among those risks, the document points in particular to the possibility of major technology firms culturally appropriating local expressions, leading to decontextualization and misrepresentation, especially when A.I. systems draw on material from Indigenous or marginalized communities.

The erosion of creativity, skills and freedom

A.I. carries a dual potential: it can empower individuals and communities by democratizing access to tools, knowledge and creative production, but it can also weaken them through overreliance, diminished human creativity and cognitive capacities, surveillance, dependency and the concentration of power. A central tension remains between the promise of democratization and the new asymmetries that can emerge through the same processes. This is especially evident in the cultural field, where algorithmic power increasingly shapes what becomes visible, valued and circulated. Recommender systems and ranking mechanisms tend to favor commercially viable content, while experimental, culturally specific or less marketable voices risk being marginalized. The document, however, also highlights community-led initiatives that show how A.I. can instead become a tool for strengthening cultural sovereignty when local communities control both knowledge and digital infrastructure. In Aotearoa New Zealand, for instance, Te Hiku Media has developed Māori language A.I. tools to support Indigenous knowledge and data governance, while India’s Bhashini initiative offers real-time translation across 22 official languages, promoting linguistic equity, cultural identity and digital inclusion at scale.

The environmental impact of A.I.

Digital technologies, including servers, blockchain, streaming services and gaming, are increasingly recognized as contributing to environmental degradation because of high energy consumption and resource extraction. A.I. in particular raises major environmental concerns, from energy consumption and carbon emissions to e-waste, requiring collaboration across technology, culture, heritage, policy and environmental science. The International Energy Agency projected in 2024 that global data center electricity demand will more than double by 2026, driven largely by A.I. adoption.

To mitigate A.I.’s environmental footprint, UNESCO calls on artists, cultural professionals and cultural institutions to pioneer sustainable approaches that prioritize renewable energy, optimized computation and critical ecological awareness through their creations. One example mentioned is Refik Anadol’s 2024 Large Nature Model, trained on more than 100 million ecosystem images, which addresses A.I.’s energy impact through partnerships with Google and NVIDIA using 100 percent renewable energy, real-time energy monitoring and open-source tools, combining scientific data with eco-conscious infrastructure. The UNESCO document identifies three key strategies addressing this specific concern: algorithmic frugality to reduce computational demands, epistemic justice to center Indigenous and Global South knowledge and symbiotic governance that links ethical goals across sectors, already visible in scalable initiatives such as African Parks, Breeze Technologies and Digital Green.

The opportunities inherent in A.I.

After mapping the risks, UNESCO identified areas where A.I. can strengthen cultural ecosystems, particularly through rapidly advancing technologies that are reshaping how culture and knowledge are accessed, produced and circulated. As the report points out, A.I. is already transforming the creative economy’s entire value chain, from initial creation and production to distribution, access and audience engagement. The group identified key strategic opportunities in:

The expansion of creativity and access

By lowering entry barriers, A.I. is enabling new creative forms of expression, idea generation and accelerated creative processes. This can also advance more equitable learning opportunities, with museums and cultural institutions cultivating A.I. literacy, creativity and critical reflection. To ensure equal opportunity for all, UNESCO suggests integrating A.I. competence into artistic curricula: while human creativity must remain essential, A.I. literacy and an informed understanding of the evolving cultural landscape are now equally important complements to it.

The protection of cultural heritage and preservation of endangered languages

A.I. can also contribute to the digital preservation of both tangible and, especially, intangible heritage through virtual reconstruction, data analysis and the transcription and documentation of endangered oral traditions and performance practices. A.I.-driven modeling and simulation tools are playing an increasingly important role in the protection of heritage sites, supporting informed conservation and protection strategies while also potentially contributing to sustainable tourism through content generation, digital storytelling and climate adaptation planning.

The E.U.-funded PERCEIVE project, mentioned as an example, brought together 12 major European museums to test A.I.’s role in conservation, including digitally enhancing the faded colors in Edvard Munch’s The Scream. A.I. has also been used to reconstruct lost works, such as three Gustav Klimt paintings destroyed during World War II, digitally reviving their colors and compositions through models trained on the artist’s existing oeuvre.

A.I. can also play a role in the protection of heritage under emergency conditions—such as after conflict or natural disasters—and strengthen efforts to combat cultural property trafficking by enabling faster identification and recovery of stolen objects. Machine learning applied to satellite imagery can also help detect looting and illicit excavation at archaeological sites, with platforms such as HeritageWatch.AI using pattern recognition and predictive analytics to support rapid heritage protection. These tools, however, require safeguards against false positives, false negatives and cultural misclassification.

Interdisciplinary collaborations, automation, efficiency and competitiveness

The use of A.I. brings technology closer to artists and creators, fostering interdisciplinary collaboration and connecting the cultural sector with technology stakeholders, opening new forms of shared creation and experimentation. Cultural and creative professionals increasingly need hybrid skills that combine artistic knowledge, technological literacy and adaptability. This means using A.I. not only in creative production, but also in administration, audience engagement and public education through technology-enabled cultural experiences. The automation of routine tasks reduces costs and enhances the competitiveness of small enterprises in the creative and cultural industries. The key tension remains between automation and agency—how to balance the scalability and efficiency enabled by A.I. with the preservation of human creativity and the inclusion of diverse cultural voices.

Epistemic justice for a shared cultural commons

After digital technologies and the internet expanded beyond specialized institutions in the late 1990s, they quickly became spaces for cultural expression, representation, communication and knowledge exchange. Yet over the past two decades, policy, industry attention and investment have largely prioritized the economic value of data as a resource to be monetized, rather than recognizing its cultural significance. As everyday culture becomes increasingly datafied, with devices filtering our experiences and channeling most of our expressions and interactions, the digital traces of daily life are already becoming what the document describes as our “digital cultural heritage”: records that represent the historical moment we live in, much as other artifacts have done in the past.

This includes both “Explicit Cultural Expressions,” meaning cultural outputs that arise from conscious, intentional acts of creation and whose cultural value is established at the time of creation, in intentional contexts or at the moment of recognition, and what the document describes as “Implicit and Latent Cultural Expressions,” whose cultural significance emerges when aggregated, forming new, large-scale cultural expressions and reflecting collective patterns.

It is particularly interesting how the analysis links the latter to the point at which A.I. systems mobilize and organize vast amounts of such data to train and develop models, contextualizing and creating meaning as they proceed. In the age of artificial intelligence, in other words, the cultural dimension is shaped by large-scale aggregation, analysis and computational training, which give rise to the cognitive capabilities of machine systems.

While this opens up a series of challenges for existing copyright laws seeking to protect the rights of cultural creators—particularly in relation to Explicit Cultural Expressions—it’s interesting to look at A.I. as a form of collective intelligence shaped by decades of digitally mediated social interaction and cultural production. This also offers a more direct awareness of the inherently collective dimension of certain kinds of cultural data—patterns, norms and values—often identified as “intangible heritage” but largely unacknowledged in law and policy, and therefore unprotected.

The latent dimension of cultural expression is therefore essential for safeguarding cultural heritage in the A.I. era, where large datasets can both preserve and transform cultural patterns over time. If artificial intelligence encourages us to think in terms of “collective commons” rather than ownership and authorship, it also requires more inclusive governance frameworks that protect recognition, ensure equitable and fair access and enable benefit-sharing among contributing communities.

In a world where access to A.I. tools is often mediated—or controlled—by a new technocracy or “techno-feudalism,” as the document puts it, made up mostly of Western-based big tech companies, creators from the Global South, Indigenous groups and other marginalized communities risk remaining at a structural disadvantage. According to UNESCO, public institutions and policymakers have a critical role to play in creating an enabling and more equitable A.I. operating environment through investments in open-source systems, public A.I. infrastructure and capacity-building initiatives. Recognizing the cognitive value of cultural data also creates opportunities to design new fiscal approaches that more equitably redistribute A.I.-generated benefits, ensuring private use of shared cultural patterns is balanced by collective returns.

In this context, UNESCO identifies “epistemic justice” as a key principle to secure both social equality and environmental awareness: respecting each community’s right to generate, legitimize and value its own knowledge through new technological tools. A project such as Heritage on the Edge, for instance, combines this at different levels, showing how A.I. can support culturally grounded conservation at climate-threatened sites, from energy-efficient 3D modeling of Rapa Nui’s moai to restoration work informed by historical texts in Bangladesh. These initiatives center local and Indigenous knowledge, data sovereignty and energy justice, positioning A.I. as a collaborative tool for sustaining culture within planetary limits.

Ultimately, this latest UNESCO document appears to acknowledge that most of the risks surrounding A.I. for culture depend on the absence of a sufficiently coordinated, enforceable and culturally grounded policy framework to regulate and direct its use. What is needed, then, are rules that safeguard and encourage more rights-based approaches to A.I., greener practices, fairer creative economies and sustainable cultural futures that continue to place human creativity and cultural rights at the center of technological development. Only under these conditions can A.I. become a tool for strengthening cultural diversity, equity, sustainability and broader cultural development, rather than a force that limits or threatens them. The central task, then, is to design and govern A.I. systems so they empower communities, protect cultural diversity and function not as instruments of control, but as infrastructures of cultural self-determination.

More in new media and digital art culture