The search for physics beyond the Standard Model remains one of the priority goals of the LHC. The lack of experimental signs and the uncertainty regarding the theory guidance heralds a new era hinting at reconsidering the current search
strategy, which is focused on classic and theoretically-founded signatures. This formerly well-motivated catalogue of highly specialized searches collapses in this new context to a set of rather arbitrary hypothesis tests. In the absence of any theory prior the traditional search strategy is far from ideal, and one would prefer a much broader search for deviations from the Standard Model. An extension to the LHC physics portfolio in form of anomaly detection is presented with the goal of maximizing the search generalizability while emphasizing automation and a reduced complexity of the data analysis. The modeling and simulation of the Standard Model background is a key aspect in this endeavour. Cutting-edge machine learning is presented as a promising tool to tackle these challenges.