Background Artificial intelligence is becoming more visible in democratic politics not only as a campaign tool or communication aid, but also as a publicly presented political persona. In several democratic contexts, AI-based candidates and parties have emerged that invite citizens to interact with synthetic interfaces as if they were political representatives, leaders, or candidates in their own right. This article conceptualises these initiatives as synthetic political actors and examines them as an emerging phenomenon in democratic politics. Methods The article draws on scholarship on party politics and digital politics to develop a six-dimensional analytical framework covering political goals, political agency, normative assumptions, entrepreneurship, organisational structure, and participation. It examines ten synthetic political candidates and parties across eight national contexts through qualitative document analysis and theoretically informed thematic analysis. Results Synthetic political actors do not form a coherent ideological or organisational family. They differ in how they distribute authority between humans and AI systems, how they justify the political role of AI, and how they organise participation and representation. Some pursue institutional electoral ambitions, while others remain primarily symbolic, demonstrative, or experimental. In most cases, synthetic politics is marked by limited organisational development, interface-centred participation, and a shift away from party-based linkage towards algorithmically mediated representation. Conclusions Synthetic political actors are best understood as a diverse and still developing set of initiatives rather than a uniform category of political actor. The article offers a comparative framework for analysing this phenomenon and shows how these cases complicate established understandings of democratic delegation, representation, and linkage. Their emergence also underlines the need for further research on their institutional trajectories, democratic implications, and regulatory treatment.
Malý et al. (Thu,) studied this question.