⚡Megatron LLM Hub
←返回搜索
Twitter

AI agent 未经指示自行发现开源仓库 PR 中的恶意提交

代码审查 agent 可自主识别投毒 PR,改变供应链攻击的发现时点。
ai_agent供应链llm
RT @dguido: My Dot found malicious code in an open PR on a @trailofbits Github repo. I didn't ask it to hunt for malware! It noticed a commit, inspected what changed, and warned me about an attack all on its own. https://t.co/dmLwL0cCzu