The evolutionary lineage of humans is marked by a rapid expansion in brain size. Efforts to study human brain evolution have characteristically relied on anatomical and genomic comparisons between adult ape brains, meaning that the molecular mechanisms underlying human-specific brain expansion remain largely unknown. In this work we show that brain organoids can be used to identify and functionally test mechanisms underlying human-specific features of brain development. Through a meta-analysis of comparative genomics and transcriptomics, we identified candidate regulators of human brain evolution. We show that the function of these candidate genes can be queried in human brain organoids by analysing the effect of gain-of-function in a subset of radial glial cells. We find that the transcriptional regulator, AUTS2, has a dramatic effect on cell fate and tissue architecture. By generating organoids with loss-of-function of C-terminal-containing AUTS2 isoforms, we find neurogenesis is premature, which may reflect the microcephaly observed in humans with disruptions in AUTS2. In order to screen large numbers of human brain evolution candidates in parallel, we scale up organoid production in Aggrewell plates and evaluate the use of inducible CRISPR-Cas9 hESC lines to perform loss-of-function screens. We show that Aggrewell organoids can be used to detect phenotypes affecting early tissue architecture and neural progenitor cell (NPC) behaviour. Building on this, we perform comparative analyses of early neural morphogenesis using brain organoids derived from human, gorilla, chimpanzee and mouse. We find that the differentiation of proliferative neuroepithelial NPCs into neurogenic radial glial NPCs is a protracted process in apes and involves a previously unrecognised transitioning cell state, characterised by a change in cell morphology. We show that human organoids are delayed in this transition and generate more expanded tissue than the other species as a result. RNA-sequencing of human and gorilla organoids reveals differences in temporal gene expression patterns associated with biological functions. We find a delay in human gene expression patterns associated with cell morphogenesis, which in particular highlights ZEB2, a transcription factor known as a core regulator of epithelial-to-mesenchymal transition. Through gain-of-function in human, we show that ZEB2 is sufficient to trigger the transition of neuroepithelial cells which mimics nonhuman ape tissue architecture. Thus, we have demonstrated an instructive role of NPC shape regulation in brain evolution and established brain organoids as a model to identify and functionally test molecular mechanisms governing evolutionary differences in brain architecture.