Thought Experiment #4: Veneta Androva. The Secret Life of Mushrooms (VIDEO)
curated by Aksiniya Peycheva
21.12.2023 - 10.01.2024
In the project “The Secret Life of Mushrooms (VIDEO)”, Veneta Androva and Tsvetomila Mihailova direct their attention towards disinformation and deep fakes – prevalent attributes of today’s media landscape. Stemming from two different fields – Veneta in visual art and Tsvetomila in computational linguistics, in the context of “Thought Experiment” project, they remotely collaborate using the Python programming language, focusing on distinct models for identifying computer-generated content, and thereby, combating mass disinformation.
Veneta’s curiosity delves into a phenomenon specific to Bulgaria – referred to as “mushroom” websites. These deceptive pages mimic typical news sites but function solely to spread propaganda and disinformation. Termed ‘mushroom’ sites due to their sporadic appearance, multiplication during active campaigns, and disappearance or inactivity post-campaign, these sites are spawned by the ‘Mushroom Machine.’ Over 370,000 anonymous sites with near-identical designs and content disseminate disinformation. They converge under four primary domains—dnes24.eu, zbox7.eu, bgvest.eu, allbg.eu—cloning into subdomains (e.g., novini24.allbg.eu). [1]
Understanding their operational processes becomes intriguing for Veneta. The underlying principle unravels as follows—on established platforms like http://share4pay.com/, any user can register using just an email address. Veneta promptly signs up, selects a site name, and within 30 minutes, a personalized ‘mushroom’ website goes live. Shielded by Cloudflare for cybersecurity, these sites remain untraceable, but the primary platform share4pay.com hosting is registered in Bulgaria. After that, an avalanche of links from the new website begins—they are present as posts on social media, forwarded to friends, acquaintances, posted in groups, etc. In this way, profits are generated on the platform, which, at some point, are paid out to the user—i.e., the main income comes from reposting the links. On the other hand, the people behind the platform earn from the ads that are everywhere on the site. Most of the news’ content is highly conservative, mildly propagandistic, mostly pro-Russian, not always fake, but certainly intended to tilt attention to certain topics and, through mass distribution, create the impression that these are ‘the news of the day.’ Processes that are extremely dangerous for society and can be exploited in any information aspect.
In her work, Veneta specifically focuses on news related to the so-called ‘gender propaganda’ and archives all news articles containing the word ‘gender’ from her new ‘mushroom website’—including their titles and images. The video she presents is a 3D animation featuring a ‘mushroom-character’ reciting all these clickbait headlines. The exhibition title, crafted in the style of clickbait rhetoric, ends ironically with (VIDEO), functioning as a typical element to attract attention and also as a nod to the artistic medium within the exhibition. Both headlines and images on the website undergo testing (with the help of Tsvetomila) using various LLM models*. These models detect whether the texts and images are generated by artificial intelligence. These tests visualize what the code ‘looks at’ and recognizes as fake, and they are incorporated into the animated video work. The second video in the exhibition focuses on an interview with a person who earns extra income from mushroom websites. His image has been anonymized. This video offers a unique insight into the scheme and its operation. In this manner, Veneta Androva questions not only the technological but also the purely human mechanisms for dealing with misinformation, the risks, vulnerabilities in the digital environment, and manipulation. In her work, these social phenomena are interpreted not solely as a function of technological progress but also as opposition to values, akin to an anti-utopia.
*Technologies based on artificial intelligence, or in particular machine learning, for recognizing deepfakes are developing dynamically, in an eternal race with the development of basic models such as BERT, GPT or Stable Diffusion. Each deepfake detection model works conditionally, with one known inaccuracy. In their experiment, Veneta and Tsvetomila decided to investigate the pattern (or what exactly it focuses on when recognizing false or misleading information). They do this using the models bert-deepfake-bg for recognizing deepfakes and bert-desinform-bg for recognizing disinformation, pre-trained specifically in Bulgarian and made available for general use on the HuggingFace platform by The Big Data for Smart Society Institute (GATE ) in Bulgaria.
[1] https://sensika.com/blog/disinformation/what-do-we-call-mushroom-websites
–
Veneta Androva graduated History of Art and Philosophy from Humboldt University Berlin and also obtained a degree in Fine Arts/Sculpture from Weißensee Academy of Art Berlin. In her artistic practise she is combining diverse media and sources such as archival, documentary or computer-generated material and painting, all related through animation in simulated environments. She took part in numerous shows in Germany and Bulgaria, as well as Austria, Argentina, Brazil, Poland, Czech Republic, Spain and in Israel, where she did part of her study at the Bezalel Academy of Arts and Design in 2016-2017. Androva is also the recipient of several scholarships, such as Elsa-Neumann-Scholarship, Artist Scholarship from Cusanuswerk, Mart Stam Scholarship and was nominated in 2020 with her work ‘From My Desert’ for the German Short Film Award. She received Prix Ars Electronica 2021 – Award of Distinction for her film ‘AIVA’ in category ‘Computer Animation’, as well as Golden Horseman for Animated Film and LUCA Gender Diversity Award at Filmfest Dresden. Her short films have been selected to screen at numerous international film festivals, including: International Leipzig Festival for Documentary and Animated Film; ARS ELECTRONICA – Festival for Art, Technology and Society; MONSTRA Lisbon Animated Film Festival; FILE Electronic Language International Festival; European Media Art Festival (EMAF); Kassel Documentary Film and Video Festival, Go Short – International Short Film Festival Nijmegen, Sofia International Film Festival, Short Waves Festival, 25FPS International Film and Video Festival, Short Film Festival Hamburg.
–
Tsvetomila Mihaylova is a Postdoctoral Researcher in Machine Learning and Robotics at Aalto University in Espoo, Finland. She has a PhD in Machine Learning and Natural Language Processing from Instituto de Telecomunicações in Lisbon, Portugal.
Art manager: Teodora Dzherekarova
*
The project is realised with the financial support of the National Culture Fund
PROGRAMME FOR RESTORATION AND DEVELOPMENT OF PRIVATE CULTURAL ORGANISATIONS