Open-weight frontier models fabricate ~72% of AI-governance numeric answers, almost never refuse, and show inverted North/South accuracy driven largely by a proportional ±10% scoring rule and sparse high-value indicators.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CY 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Benchmarking Open-Weight Foundation Models for Global AI Technical Governance
Open-weight frontier models fabricate ~72% of AI-governance numeric answers, almost never refuse, and show inverted North/South accuracy driven largely by a proportional ±10% scoring rule and sparse high-value indicators.