I don't have any trust in these Chinese models to write code either: "CrowdStrike Research: Security Flaws in DeepSeek-Generated Code Linked to Political Triggers " [https://www.crowdstrike.com/en-us/blog/crowdstrike-researche...]