LLM 编程代理带来新安全风险
As @HeidyKhlaaf just pointed out to me, quite rightly, there is also the huge new risk of huge piles...
精选理由
Gary Marcus 和 Nathan Hamiel 说，LLM 编程代理会生成大量不安全的代码，这很危险。
Gary Marcus 和 Nathan Hamiel 在 Substack 上发文指出，使用大语言模型（LLM）的编程代理存在巨大风险，可能导致大量不安全的代码生成。他们强调，这种技术结合可能成为新的安全噩梦。
原文 · Gary Marcus
As @HeidyKhlaaf just pointed out to me, quite rightly, there is also the huge new risk of huge piles...
As @HeidyKhlaaf just pointed out to me, quite rightly, there is also the huge new risk of huge piles of insecure vibe-written code. (See my essay w @nathanhamiel , LLMs + Coding Agents = Security Nightmare, on that score: garymarcus.substack.com/p/llms-coding-… ) 💬 0 🔄 0 ❤️ 2 👀 478 📊 1 ⚡
