We aim to use first-principles calculations at finite temperatures in combination with machine learning (ML) techniques to derive an accurate picture of hydrogen embrittlement in Ni-based superalloys. The ML-based interatomic potential will allow for the determination of the temperature dependent antiphase boundary energy (APB energy) for γ' precipitates (i.e., Ni3Al) which is a measure for the ability of dislocations to enter and propagate in these precipitates. The resulting knowledge of this key quantity together with the availability of a very accurate machine trained interatomic potential will provide a fundamental basis for the future design of hydrogen-tolerant and creep resistant Ni-based superalloys.