AT 2016dah and AT 2017fyp are fairly typical Andromeda galaxy (M 31) classical novae. AT 2016dah is an almost text book example of a ‘very fast’ declining, yet uncommon, Fe ii‘b’ (broad-lined) nova, discovered during the rise to peak optical luminosity, and decaying with a smooth broken power-law light curve. AT 2017fyp is classed as a ‘fast’ nova, unusually for M 31, its early decline spectrum simultaneously shows properties of both Fe ii and He/N spectral types – a ‘hybrid’. Similarly, the light curve of AT 2017fyp has a broken power-law decline but exhibits an extended flat-topped maximum. Both novae were followed in the UV and X-ray by the Neil Gehrels Swift Observatory, but no X-ray source was detected for either nova. The pair were followed photometrically and spectroscopically into their nebular phases. The progenitor systems were not visible in archival optical data, implying that the mass donors are main-sequence stars. What makes AT 2016dah and AT 2017fyp particularly interesting is their position with respect to M 31. The pair are close on the sky but are located far from the centre of M 31, lying almost along the semiminor axis of their host. Radial velocity measurements and simulations of the M 31 nova population leads to the conclusion that both novae are members of the Andromeda Giant Stellar Stream (GSS). We find the probability of at least two M 31 novae appearing coincident with the GSS by chance is $\sim \!1{{\ \rm per\ cent}}$. Therefore, we claim that these novae arose from the GSS progenitor, not M 31 – the first confirmed novae discovered in a tidal steam.