Researchers have found that Large Language Models (LLMs) can produce convincing political text, sparking concerns about the potential for synthetic discourse to spread during crises and social unrest. As LLMs improve, existing methods for detecting AI-generated text, such as analyzing sentence structure and token patterns, may become less effective1. To address this, a new approach has been developed, focusing on the broader social and computational context in which these models operate. This perspective allows for a more nuanced understanding of how LLMs can be used to generate persuasive and potentially misleading political content. The implications of this research extend far beyond the technical realm, with significant consequences for policy, security, and the future of work. So what matters to practitioners is that they must develop more sophisticated methods for identifying and mitigating the influence of AI-generated political discourse, in order to maintain the integrity of public debate and decision-making.