A new preprint (arXiv:2605.27509, v1, May 2026) presents an inversion technique to recover the high-mass end of the globular cluster initial mass function (GCIMF) by correcting present-day globular cluster mass functions (GCMFs) for environment-dependent dynamical mass loss. The method relies on N-body simulations in time-varying tidal fields, calibrated across host halo masses from 10^9 to 10^12 solar masses, and avoids assuming any functional form for the GCIMF a priori. Applying this to both individually resolved cluster systems and statistical samples yields GCIMFs that are shifted to higher masses and follow power-law slopes at the high-mass end, with slopes steepening systematically in more massive halos. This finding challenges the long-standing view of a universal turnover mass (~2x10^5 solar masses) as purely a formation imprint, instead linking it to differential dynamical evolution modulated by galactic environment. Prior work, such as the semi-analytic models of Fall & Zhang (2001, ApJ) that assumed a universal initial power-law GCIMF evolving under two-body relaxation alone, missed the observed halo-mass dependence recovered here. Complementary constraints from JWST high-redshift cluster candidates (e.g., Forbes et al. 2024, MNRAS) suggest that the steeper GCIMFs in massive galaxies trace denser, more clustered star formation at z>2, providing an empirical bridge between local observations and early galaxy assembly. Limitations include reliance on simulation-calibrated mass-loss rates that may underrepresent stochastic tidal shocks in merging galaxies, and the sample spans only ~dozen well-studied systems plus binned statistical data rather than a complete volume-limited census. As a preprint, these results await peer review.