Tungsten (W) is considered the prime candidate for the structural material of fusion reactor components [1]. However, the accumulation of helium (He) in W caused by ion implantation leads to He segregating at the grain boundaries (GBs) and weakening the cohesive strength by screening W interaction across the GB. This is especially problematic as W GBs are more brittle than the grains [2], and He further lowers the threshold for crack formation and propagation. In this work an atomistically-informed machine learning (ML) approach is developed to parametrize traction-separation data from molecular dynamics (MD) simulations of He-inhabited W GBs. The resulting ML surrogate, which is trained by data from virtual tensile MD simulations, enables a data-driven damage model that can capture the traction-separation behavior of He covered W GBs [3, 4]. As continuum scale effects of the build-up of He in W GBs are tied to He-bubble pressure [1], the ML surrogate bridges the gap between length scales [3]. The traction-separation database generated by atomistic modeling correlates the deterioration of W GBs fracture mechanical behavior to He-coverage. From this, the resulting ML model ties the atomistic scale embrittlement mechanism of He-bubble pressure to the deterioration of the cohesive properties of W GBs on a continuum scale. The implication of this work is that ML can be used as an effective bridge between smaller length scale mechanisms to model the resulting continuum scale effects. REFERENCES [1] Linke J.,et al., “Challenges for plasma-facing components in nuclear fusion,” Matter Radiat. Extremes, vol. 4, no. 5, 2019, Art. no. 056201. [2] Gludovatz B., et al, “Influence of impurities on the fracture behaviour of tungsten”, Eng. Fracture Mechanics, vol. 91, no. 22, pp. 3006–3020, 2011. [3] Fernandez M., et al, “Application of artificial neutral networks for the prediction of interface mechanics: a study on grain boundary constitutive behaviour,” Adv. Model. Simul. Eng. Sci, vol. 7, 2020, Art. no. 1. [4] Barrows W., Dingreville R., and Spearot D., “Traction-separation relationships for hydrogen induced grain-boundary embrittlement in nickel via molecular dynamic simulation,” Mater. Sci. Eng. A, vol. 650, pp. 354–364, 2015.