LogoThread Easy
  • 探索
  • 線程創作
LogoThread Easy

Twitter 線程的一站式夥伴

© 2025 Thread Easy All Rights Reserved.

探索

Newest first — browse tweet threads

Keep on to blur preview images; turn off to show them clearly

RT @a1zhang: Happy to have done this! Definitely don’t have the same level of credentials as some of the other guests but hopefully some of…

RT @a1zhang: Happy to have done this! Definitely don’t have the same level of credentials as some of the other guests but hopefully some of…

Asst professor @MIT EECS & CSAIL (@nlp_mit). Author of https://t.co/VgyLxl0oa1 and https://t.co/ZZaSzaRaZ7 (@DSPyOSS). Prev: CS PhD @StanfordNLP. Research @Databricks.

avatar for Omar Khattab
Omar Khattab
Fri Oct 31 01:27:24
👀

👀

Wonderer. Amor fati. Scaling trust.

avatar for Michael Frank Martin
Michael Frank Martin
Fri Oct 31 01:27:11
RT @Noahpinion: Here's a longer-term chart. The simple story is correct. Capitalism created Poland's miracle, and socialism created Venezue…

RT @Noahpinion: Here's a longer-term chart. The simple story is correct. Capitalism created Poland's miracle, and socialism created Venezue…

AI Optimist. Empiricist, not 'rationalist'. Anti world government.

avatar for renji
renji
Fri Oct 31 01:25:52
博客:https://t.co/CiBFskE4fX
申请参与测试:https://t.co/20WIxCGF4Y

博客:https://t.co/CiBFskE4fX 申请参与测试:https://t.co/20WIxCGF4Y

致力于让每个想拥抱AI的人都能找到适合自己的AI产品,助力企业定制AIGC应用

avatar for AIGCLINK
AIGCLINK
Fri Oct 31 01:24:27
OpenAI也搞了款安全智能体:Aardvark,基于GPT-5的,主打一个自主挖漏洞、写补丁

目前这个工具在测试中,成功识别了92%的漏洞

Aardvark用LLM推理+工具调用像人类安全员一样,读代码,分析逻辑、编写并运行测试、用工具发现漏洞

首先分析整个代码仓库,建立一个威胁模型清单,来理解项目的安全目标,设计架构

当有新代码提交时,会针对整个代码库及威胁模型清单逐条检查,看新代码是否有漏洞,如有,会解释漏洞成因,并在相关代码处添加注释,供人工评审

发现潜在漏洞后,会在隔离的沙箱环境中尝试触发该漏洞来验证,降低误报率

漏洞被确认,会集成一个OpenAI Codex代码生成模型来自动生成修复补丁,并且这个补丁会附在漏洞报告中

目前这个工具正在内测

#AI安全 #代码安全工具 #Aardvark

OpenAI也搞了款安全智能体:Aardvark,基于GPT-5的,主打一个自主挖漏洞、写补丁 目前这个工具在测试中,成功识别了92%的漏洞 Aardvark用LLM推理+工具调用像人类安全员一样,读代码,分析逻辑、编写并运行测试、用工具发现漏洞 首先分析整个代码仓库,建立一个威胁模型清单,来理解项目的安全目标,设计架构 当有新代码提交时,会针对整个代码库及威胁模型清单逐条检查,看新代码是否有漏洞,如有,会解释漏洞成因,并在相关代码处添加注释,供人工评审 发现潜在漏洞后,会在隔离的沙箱环境中尝试触发该漏洞来验证,降低误报率 漏洞被确认,会集成一个OpenAI Codex代码生成模型来自动生成修复补丁,并且这个补丁会附在漏洞报告中 目前这个工具正在内测 #AI安全 #代码安全工具 #Aardvark

博客:https://t.co/CiBFskE4fX 申请参与测试:https://t.co/20WIxCGF4Y

avatar for AIGCLINK
AIGCLINK
Fri Oct 31 01:24:26
🤔

🤔

Wonderer. Amor fati. Scaling trust.

avatar for Michael Frank Martin
Michael Frank Martin
Fri Oct 31 01:22:16
  • Previous
  • 1
  • More pages
  • 1644
  • 1645
  • 1646
  • More pages
  • 2118
  • Next