Different diagnostics of hot star wind mass-loss rates provide results that are difficult to reconcile with each other. The widely accepted presence of clumping in hot star winds implies a significant reduction of observational mass-loss rate estimates from diagnostics that depend on the square of the density. Moreover, the ultraviolet P V resonance lines indicate a possible need for an even stronger reduction of hot star mass-loss rates, provided that P V is a dominant ionization stage of phosphorus, at least in some hot stars. The latter assumption is challenged by the possible presence of extreme ultraviolet (XUV) radiation. Here, we study the influence of XUV radiation on the P V ionization fraction in hot star winds. Using a detailed solution of the hydrodynamical radiative transfer and statistical equilibrium equations, we confirm that a sufficiently strong XUV radiation source might decrease the P V ionization fraction, possibly depreciating the P V lines as a reliable mass-loss rate indicator. However, the XUV radiation also influences the ionization fraction of heavier ions that drive the wind, leading to a decrease of the wind terminal velocity. Consequently, we conclude that XUV radiation alone cannot bring theory and observations into agreement. We fit our predicted wind mass-loss rates using a suitable formula and we compare the results with the observational mass-loss rate diagnostics. We show that for supergiants and giants the theoretical predictions do not contradict the mass-loss rate estimates based on X-ray line profiles or density-squared diagnostics. However, for main-sequence stars, the predicted mass-loss rates are still significantly higher than those inferred from P V or X-ray lines. This indicates that the 'weak wind problem' recently detected in low-luminosity main-sequence stars might also occur, to some extent, for stars with higher luminosity.