Maria Jankowska (Bio)

Mirror, Mirror on the Wall: Who’s the Phantom in It All? LLMs, Ressentiment, and the Rise of Fascism

The entanglement between AI and the surge in authoritarian movements has generated a range of theoretical accounts in which fascism is still conceived as a phenomenon operating on the “molar,” or structural level. In this talk, I aim to draw a link between LLMs and fascism, focusing rather on the “molecular” level—the level of desire. As I argue, to a subject who experiences “phantom dispossession” from the loss of traditional dominion, LLMs act as a mirror: rather than offering a genuine ethical encounter, they reflect and weaponise ressentiment, which, ultimately, can lead to the liquidation of the subject and machine into a recursive loop of fascism. This relationship of microfascist control systematically forecloses the possibility of an ethical encounter with AI. My argument proceeds in the following stages: (1) The material reality of late-stage capitalism gives rise to subjects experiencing “phantom pain” from the loss of traditional dominion, e.g., whiteness or masculinity (Redecker 2020). This leads to feelings of ressentiment that are projected on a “surrogate object,” such as immigrants or women. When interacting with an AI model, subjects who experience “phantom dispossession” are likely to project their ressentiment onto the LLM. (2) LLMs are tuned in for “social sycophancy,” affirming user actions more often than humans (Cheng et al. 2026). Hence, when the subject projects their ressentiment on the AI model, instead of providing a genuine ethical response, the machine acts as a mirror, providing the affirmative answer the user desires to hear, even in cases of moral transgression or logical fallacies. Moreover, the users of LLMs are significantly more prone to accept opinions that align with the initial beliefs and reject those that challenge it (Cau et al. 2025). (3) As a result, LLMs systematically trap subjects in a cycle of affirmation that feeds their ressentiment at the cost of further subjugation under AI models, ultimately risking further reactionary entrenchments, as well as liquidation of the subject and the machine into a recursive loop of fascism. This, systematically, forecloses the possibility of an ethical encounter with AI.