Spray and Pray is a speculative animated documentary about “mushroom websites” — automated news networks built to monetize disinformation through advertising and platform circulation.

The film follows the human–machine arrangements behind these infrastructures: link-spreading micro-labor, domain operators running large clusters of sites, ad-tech intermediaries, and algorithmic systems that amplify reach. Developed through long-term research on Bulgarian-language disinformation, the project combines CGI animation with synthetic voices and machine-learning assisted analysis as both method and material. Spray and Pray renders an opaque economy of influence perceptible — tracing how agency and responsibility circulate within automated systems.

The project was developed through a research-based artistic process combining media monitoring, investigative sources, and collaboration with machine-learning researchers. Rather than focusing on individual cases, the research examined patterns, infrastructures, and circulation mechanisms of disinformation networks, with particular attention to Bulgarian-language content — an area that remains significantly underrepresented in AI-based detection systems despite being actively targeted by coordinated influence operations.

The research began in 2023 and combined investigative methods, media monitoring, and academic sources with collaboration with machine-learning researchers. In collaboration with computational linguist Tsvetomila Mihaylova and with support from the GATE Institute (Sofia), Bulgarian-language machine-learning models were used to analyse collected headlines and articles. These experiments explored machine-generated text detection, disinformation likelihood, and sentiment analysis. Rather than using these systems as tools for verification, the project examined their limitations, biases, and interpretative ambiguities.

The film adopts a hybrid documentary form in which research material is translated into speculative cinematic situations. Characters are constructed from anonymized documentary sources and collected narratives, while voice performances are constructed through neural voice synthesis using recorded performances and contributed voice material. This deliberate ambiguity between documentary and constructed elements reflects the logic of contemporary disinformation environments, where boundaries between authentic and fabricated content become increasingly unstable.

By combining documentary research, CGI animation, and machine-learning experiments, Spray and Pray uses speculation as an investigative method to render otherwise invisible infrastructures perceptible.