https://www.ithome.com.tw/news/168417
OWASP十大LLM應用程式風險
- LLM01:2025 提示詞注入(Prompt Injection)
- LLM02:2025 敏感資訊揭露(Sensitive Information Disclosure)
- LLM03:2025 供應鏈風險(Supply Chain)
- LLM04:2025 資料與模型投毒(Data and Model Poisoning)
- LLM05:2025 不當輸出處理(Improper Output Handling)
- LLM06:2025 過度代理授權(Excessive Agency)
- LLM07:2025 系統提示詞洩露(System Prompt Leakage)
- LLM08:2025 向量與嵌入弱點(Vector and Embedding Weaknesses)
- LLM09:2025 錯誤資訊(Misinformation)
- LLM10:2025 無限資源耗盡(Unbounded Consumption)