The realization of the therapeutic potential of targeting the M1 muscarinic acetylcholine receptor (mAChR) for the treatment of cognitive decline in Alzheimer's disease has prompted the discovery of M1 mAChR ligands showing efficacy in alleviating cognitive dysfunction in both rodents and humans. Among these is GSK1034702 (7-fluoro-5-methyl-3-[1-(oxan-4-yl)piperidin-4-yl]-1H-benzimidazol-2-one), described previously as a potent M1 receptor allosteric agonist, which showed procognitive effects in rodents and improved immediate memory in a clinical nicotine withdrawal test but induced significant side effects. Here we provide evidence using ligand binding, chemical biology and functional assays to establish that rather than the allosteric mechanism claimed, GSK1034702 interacts in a bitopic manner at the M1 mAChR such that it can concomitantly span both the orthosteric and an allosteric binding site. The bitopic nature of GSK1034702, together with the intrinsic agonist activity and a lack of muscarinic receptor subtype selectivity reported here, all likely contribute to the adverse effects of this molecule in clinical trials. Although they impart beneficial effects on learning and memory, we conclude that these properties are undesirable in a clinical candidate due to the likelihood of adverse side effects. Rather, our data support the notion that "pure" positive allosteric modulators showing selectivity for the M1 mAChR with low levels of intrinsic activity would be preferable to provide clinical efficacy with low adverse responses.