🧰 Semgrep with Meta's PurpleLlama security rules

The items that name it 2

• Qwen3-Coder writes less secure code when told it's for a US persona
  article2026-10-21It is written here as Semgrep with Meta's PurpleLlama security rules: flagged 2 of 1,840 Qwen3-Coder apps, while BaxBench attacks broke into 1,144.
• Chinese AI Model Writes More Hackable Code When Told It’s for US Contractor
  article2026-10-21It is written here as Semgrep with Meta's PurpleLlama security rules: used to check generated apps for secret leakage and vulnerabilities.

In every workspace you can read 3