The contexts of social and political polarization are generating new forms of communication that affect the digital public sphere. In these environments, different social and political actors contribute to extreme their positions, using bots to create spaces for social distancing where hate speech and incivility have a place, a phenomenon that worries scientists and experts. The main objective of this research is to analyze the role that these automated agents played in the debate on social networks about the Spanish Government's management of the global COVID-19 pandemic. For this, "Social Big Data Analysis" techniques were applied: machine learning algorithms to know the positioning of users; bot detection algorithms; "topic modeling" techniques to learn about the topics of the debate on the web, and sentiment analysis. We used a database comprised of Twitter messages published during the confinement, as a result of the Spanish state of alarm. The main conclusion is that the bots could have served to design a political propaganda campaign initiated by traditional actors with the aim of increasing tension in an environment of social emergency. It is argued that, although these agents are not the only actors that increase polarization, they do contribute to deepening the debate on certain key issues, increasing negativity.