Social bots are increasingly deployed to sway public opinion during high‑stakes events. We analyze 42 million election‑related tweets from X (formerly Twitter), label 4.3 million accounts with BotMoE, and compare the temporal behavior, topical focus, and network position of bots versus human users. Bots generate five to seven times more hashtags and three times more tweets per peak month than humans, with spikes around the Trump assassination attempt, party conventions, and Biden’s withdrawal. Topic modeling shows bots consistently amplifying pro‑Trump slogans and media‑driven tags (e.g., #foxnews), whereas human discourse is more diverse and fades quickly. Removing bots shatters the reply network from 5,580 to 32,700 Louvain communities, revealing that automated accounts act as structural bridges and inflate apparent cross‑group cohesion. These findings indicate that a relatively small bot population can reshape both the content and the topology of political conversation, underscoring the need for robust detection and mitigation before future elections.