arXiv cs.AITuesday · May 26, 2026FREE

A governance horizon for ethical-use constraints in open-weight AI models

A study published on arXiv (2605.24383) audited 2,142,823 model repositories on Hugging Face Hub to test whether disclosure-based governance can sustain traceability across deep model lineages. The researchers found that restriction evidence decays with a half-life of 1.31 derivation steps (R²=0.98). Beyond seven downstream generations, at least 80% of descendant models lack sufficient public evidence for a governance determination, a depth boundary formalized as the governance horizon. Platform-level interventions to restore missing license metadata reveal that policy design—not enforcement alone—is the binding factor. Inheritance-only designs require near-complete enforcement to move the horizon, whereas a mandatory-declaration design that explicitly resolves orphan lineage components shifts the horizon already at moderate enforcement. The structural bottleneck is lineages with no metadata.