Hooded Language models (HLM) Rugcheck — Risk Score 4/10