Selected Papers of #AoIR2024: The 25th Annual Conference of the Association of Internet Researchers Sheffield, UK / 30 Oct - 2 Nov 2024 Suggested Citation (APA): Spaggiari, C., Gemini, L., Brilli, S., (2024, October). Generic War Imaginaries: AI-Generated Images of The Israel-Gaza Conflict In The Adobe Stock Controversy. Paper presented at AoIR2024: The 25th Annual Conference of the Association of Internet Researchers. Sheffield, UK: AoIR. Retrieved from http://spir.aoir.org. GENERIC WAR IMAGINARIES: AI-GENERATED IMAGES OF THE ISRAEL-GAZA CONFLICT IN THE ADOBE STOCK CONTROVERSY Chiara Spaggiari University of Urbino Carlo Bo Laura Gemini University of Urbino Carlo Bo Stefano Brilli University of Urbino Carlo Bo Introduction This paper explores the evolving documentality of AI-generated images and their role in the industrial production of war imaginaries. The focus of our analysis is the paradigmatic case of Adobe Stock selling photorealistic AI-generated images depicting the Israel-Gaza conflict. Following the outbreak of the war in Gaza on October 7, 2023, Adobe Stock began hosting AI-generated images created by users that represented the conflict. These images, available for paid download, were used in both online and print media. A public debate arose when news outlets published these images, triggering criticism over the lack of proper disclosure regarding their AI origins. Our research question is therefore: Whether and how AI-generated images are used to document specific war event as the war in Gaza? Theoretical Background This study examines the concept of documentality, a quality traditionally associated with photography for its ability to offer referential and indexical representations. This has been particularly significant within the informational system, where images serve as evidence and testimony in the representation of news and events (Becker 1995; Coleman 1998). Scholars have explored how documentary images within photojournalism and photography are historically shaped by the codes of the information system, which rely on agreements, practices, and assumptions shared between photographers and news organizations (Schwarz 2003; Carlson 2009; Wahl-Jorgensen & Carlson 2021). However, the rise of AI-generated images, including those associated with fake news and deepfakes, has intensified debates over the authenticity and reliability of photographic objects, challenging the stability of documentary codes (Lister 2013; Lehmuskallio et al. 2019; Thomson, Ryan & Matich 2024). This shift aligns with broader theoretical discussions within media studies regarding the “crisis of photography.” This crisis, amplified by the algorithmic turn (Uricchio 2011), has been identified as part of the post-photographic turn (Dewdney 2021; Zylinska 2022; Grazioli 2024; McQuire et al. 2024). Generative visual media (Arielli & Manovich 2023) have pushed these discussions to their limits, particularly questioning the indexicality of photography and its ability to represent dramatic events, such as in war photojournalism. In addition, generative visual media intertwine with the processes of datafication and platformization of images (Anderson 2017; Taffel 2021), serving the visual content industry’s needs. Stock agencies, as key players in this industry, leverage digital technologies to meet demands for generic aesthetics and cost-effective image production, increasingly shaping contemporary visual culture and news illustration practices (Blaschke 2014, 2019; Szendy 2020; Thurlow et al. 2020; Aiello et al. 2023). Methodology A key challenge of this study was to avoid treating documentality as an inherent property of an image. Instead, we conceptualize it as a category shaped by multiple actors. To this end, we adopted Gillian Rose’s (2016) framework, analyzing images through four sites where meaning is constructed: • Production site: We examined how Adobe Stock and other platforms (iStock, Shutterstock, Unsplash, Pixabay, Pexels, 123RF, Alamy) label and regulate AI- generated images. This included analyzing policies, classification systems, and rules for indexing and uploading content. • Audiencing site: In the first phase, we collected 57 articles from Google News in February 2024 related to the Adobe controversy. Using a qualitative content analysis and a reference codebook, we examined these texts. In the second phase, we are conducting interviews with 15 qualified experts, including photo editors, photojournalists, and documentary photographers, to assess the documentary potential of AI-generated images. • Circulation site: We conducted a framing analysis of all online occurrences (97) of six “widely shared” images, as identified in news coverage of the controversy, using Google SERP and Lens APIs. • Image site: A computational analysis was performed on the RGB spectra of 762 most downloaded images from Adobe Stock, collected in October 2024. This aimed to detect technical and formal continuities or distinctions between photographic and AI-generated images. Preliminary Results Initial findings reveal significant ambiguities in the documentality of photorealistic AI- generated images depicting specific events. These insights are organized by site: • Production site: Adobe Stock does not classify AI-generated images separately. Instead, they are included under “photos” or “illustrations” based on visual resemblance to photographic norms. Contributors must label AI-generated images and ensure that depicted people or properties are fictional unless they provide signed releases. References to real events are prohibited in titles but allowed in keywords, enabling subtle connections to actual events. Adobe’s approach contrasts with stricter policies on other platforms and aligns with its focus on photo-editing tools. • Audiencing site: The controversy revolved around the use of AI-generated images without clear labeling, raising concerns about misinformation. Adobe faced criticism for transparency issues but highlighted its role in the Content Authenticity Initiative, a cross-industry effort to verify digital media provenance. While some criticized Adobe’s policies as ethical failings, others viewed them as necessary for market competitiveness, emphasizing shared responsibility for image regulation. • Circulation site: The six “widely shared” AI-generated images circulated mostly in articles criticizing their use. In the other articles, they were often accompanied by captions referencing general war and destruction rather than specific events, with their AI origin rarely disclosed. These images functioned more as illustrative elements than as documentary evidence, consistent with stock photo conventions. • Image site: Computational analysis, dividing the dataset into 10 clusters, revealed no clear separation between photographic and AI-generated images. However, AI-generated images tended to appear in clusters with darker color tones, suggesting a preference for dramatic, painterly effects rather than adherence to natural light. Conclusion AI-generated images occupy an intermediate space between their potential to document specific events (as observed in the production site and from the information system) and their tendency toward generic, fictionalized representations, even in dramatic contexts (as seen in the circulation site and the images themselves). Future research will focus on analyzing the circulation patterns of the 762 images from Adobe Stock, using Google SERP API to observe their contextualization as documentary or illustrative content. Additionally, for the image site, we will further investigate the aesthetic and iconographic continuities between this photographic and photorealistic AI-generated images to assess whether the latter develop distinct visual characteristics. References Aiello, G., Thurlow, C., Portmann, L., (2023), Desocializing Social Media: The Visual and Media Ideologies of Stock Photography, in Social Media + Society, January-March: 1–14. Anderson, Steve F. (2017), Technologies of Vision: The War Between Data and Images. Cambridge, MA: The MIT Press. anything else?). In J. Zylinska (Ed.), The future of media (pp. 315-332). London: Goldsmiths Press. Arielli, T., Manovich, L., (2023), Artificial Aesthetics: Generative AI, Art, and Visual Media, manovich.net. Becker, S. H. (1995). Visual sociology, documentary photography, and photojournalism: It’s (almost) all a matter of context. In J. Prosser (Ed.), Image-based research: A sourcebook for qualitative researchers (pp. 74-85). Routledge. Blaschke, E., (2019), From Microform to the Drawing Bot: The Photographic Image as Data, in Grey Room 75, Spring 2019, pp. 60–83 Blashcke, E., (2014), Making History a Slightly Profitable Thing”: The Bettmann Archive and the Commodification of Images, Visual Resources: An International Journal of Documentation, 30:3, 222-238 Boccia Artieri, G., & Gemini, L. (2019). Mass media and the web in the light of Luhmann’s media system. Current Sociology, 67(4), 563-578. Carlson, M. (2009). THE REALITY OF A FAKE IMAGE News norms, photojournalistic craft, and Brian Walski’s fabricated photograph. Journalism Practice, 3(2), 125–139. Coleman, A. D. (1998). Depth of field: Essays on photography, lens culture, and mass media. University of New Mexico Press. Dewdney, A. (2021). Forget photography. Goldsmiths Press. Frosh, P. (2020). Is commercial photography a public evil? Beyond the critique of stock photography. In Photography and its publics (pp. 187–206). Bloomsbury. Frosh, P., (2003), The image factory: Consumer culture, photography and the visual content industry. Berg. Frosh, P., (2008), “Stock Photography” in The International Encyclopedia of Communication, John Wiley & Sons. Grazioli, E. (2024). Dimensioni. In B. Grespi & F. Villa (Eds.), Il postfotografico. Dal selfie alla fotogrammetria digitale (pp. 87-98). Einaudi, Torino. Lehmuskallio, A., Häkkinen, J., Seppänen, J., (2018), Photorealistic computer- generated images are difficult to distinguish from digital photographs: a case study with professional photographers and photo-editors, in Visual Communication, Vol. 0(0) 1–25. Lister, M., (2013), The photographic image in digital culture. London: Routledge. MacKenzie, A., Munster, A., (2019), Platform Seeing: Image Ensembles and Their Invisualities, in Theory, Culture & Society, Vol. 36(5), 3–22. McQuire, S., Pfefferkorn, J., Sunde, E. K., Lury, C., & Palmer, D. (2024). Seeing photographically. Media Theory, 8(1), 01-18. Rettberg, J., (2023), Machine Vision: How Algorithms Are Changing the Way We See the World, Polity Press. Rose, G., (2016), Visual Methodologies, Sage. Schwartz, D., (2003), Professional Oversight: Policing the Credibility of Photojournalism, in Image Ethics in the digital age, University of Minnesota Press, pp 27-52 Szendy, P., (2019), The Supermarket of the Visible. Toward a General Economy of Images, Fordham University Press Taffel, S., (2020), Google’s lens: computational photography and platform capitalism, in Media, Culture & Society 2021, Vol. 43(2) 237–255 Thurlow, C., Aiello, G., & Portmann, L. (2020). Visualizing teens and technology: A social semiotic analysis of stock photography and news media imagery. New Media & Society, 22(3), 528–549. Uricchio, W. (2012). The algorithmic turn: Photosynth, augmented reality and the changing implications of the image. In G. Bolin (Ed.), Cultural technologies: The shaping of culture in media and society (pp. 19-35). Routledge. Wahl-Jorgensen, K., & Carlson, M. (2021). Conjecturing fearful futures: journalistic discourses on deepfakes. Journalism Practice, 15(6), 803-820. Zelizer, B., (2010), About to die: How News Images Move the Public, Oxford University press Zucconi, F., (2018), Displacing Caravaggio: Art, Media, and Humanitarian Visual Culture, Palgrave Macmillan. Zylinska, J. (2022). The future of photography. Does photography have a future? (Does