The adoption of hydrogen as an alternative fuel in aircraft has the potential to reduce the climate effect of aviation significantly. However, hydrogen leakage during production, storage, or use can offset these benefits by altering atmospheric chemistry and composition, particularly through interactions with methane, ozone, and stratospheric water vapour. This study employs surrogate models based on recurrent and convolutional neural networks to simulate the climate effects of hydrogen leaks, achieving rapid projections 30,000 times faster than conventional climate models, with an error margin of less than 5%. This efficiency enables the quantification of uncertainties related to hydrogen leakage rates and atmospheric chemistry through Monte Carlo simulations, allowing for an assessment of their contributions to radiative forcing under various Shared Socioeconomic Pathway (SSP) climate scenarios. By 2100, the radiative forcing from aviation-induced hydrogen leaks is projected to reach 43.9 ± 21.2 mW m-2 (±1σ) under the stringent climate change mitigation projection SSP1-2.6, accounting for 35% of aviation’s total radiative forcing. Under the more conservative scenario SSP3-7.0, higher methane levels reduce the oxidative capacity of the atmosphere, lowering the projection to 17.7 ± 6.9 mW m-2 (±1σ) which corresponds to approximately 5% of the aviation’s total radiative forcing for this scenario. These findings demonstrate that hydrogen leaks have the potential to substantially contribute to aviation’s total radiative forcing, with the magnitude of their impact heavily influenced by background climate conditions. It will be important to minimise hydrogen leakage to fully harness the climate benefits of transitioning to hydrogen as a fuel for aircraft.