●builderBe cautious when using models with heavy geopolitical fine-tuning for security-critical coding tasks.
●researcherYou should investigate how alignment tuning affects downstream code safety.
●policyThis highlights risks in managing the security profiles of internationally developed open-weight models.