开启时会模糊预览图，关闭后正常显示
在外部合作企业与开源项目中验证结果； 识别率达到 92%（针对已知与合成漏洞） 多个漏洞获得 CVE 编号 其分析深度被合作伙伴评价为“可与人类资深安全研究员媲美”。 Aardvark 的出现标志着软件安全从“被动检测”进入“智能防御自动化”阶段。 详细介绍：https://t.co/Yp87I29IEY
学AI找小互，找小互，上 https://t.co/4PVaHEr5r3 ...
兄弟们 这个厉害了 OpenAI 推出 AI 安全Agent： Aardvark 它可以24 小时不停地“巡逻”代码库帮你发现漏洞和bug 更牛逼的是，它不仅发现还能自动在沙盒里帮你测试这个漏洞或者bug 然后自动帮你修复，并给你提交一个报告😂 相当于你雇佣了一个24小时的网络安全员... Aardvark 已在 OpenAI 内部代码库中运行数月，它能识别超过 9 成的已知问题 它的工作方式大致是： - 读代码 → 看懂你的程序逻辑； - 找问题 → 在每次提交代码后检查可能的漏洞； - 试漏洞 → 自己在安全环境里尝试利用漏洞，验证真伪； - 修漏洞 → 用 Codex 自动写出修复方案并提交补丁。 为什么重要？ - 普通企业很难雇到足够的安全专家 - 每年都有数万漏洞被公开； - Aardvark 能 24 小时不停地“巡逻”代码库，提前发现问题 - 它不仅找安全漏洞，还能发现逻辑错误或隐私风险并自动修复
在外部合作企业与开源项目中验证结果； 识别率达到 92%（针对已知与合成漏洞） 多个漏洞获得 CVE 编号 其分析深度被合作伙伴评价为“可与人类资深安全研究员媲美”。 Aardvark 的出现标志着软件安全从“被动检测”进入“智能防御自动化”阶段。 详细介绍：https://t.co/Yp87I29IEY
One of the absolute highlights of my year has been getting to know and become friends with @danielkoeth. In addition to being one of the most brilliant and creative people I know, he is a wonderful person through and through -- a total mensch. We took a conviction bet on each other back in the spring, and since then, he's literally transformed the way I see the sky. I didn't even know it was possible for art to do that. And that's not all: in the past six months, his work and style have leveled up over and over again. Just like the best haiku find uniqueness of image in a tiny amount of space and structure, Daniel is a master of *duality* of image: someone is flying at the same instant as they are falling, or pulling back the curtain to the sky that very slightly possibly might actually be the theater's back wall. Whereas in many of Daniel's previous works, these dualities zigzagged along different vectors at once, his new series hones and focuses the effect into single coherent bursts around a single subject in a way that keeps the emotion of the image alive in my mind for hours. Of course, mensch that he is, Daniel has also been a tremendous supporter of me and my work, as I've been discovering my own art practice -- and I hear the same from many others in the space. And this particular work of his is so deeply personal to me -- he literally crafted a lifelong puzzle as a tribute to our connection. Blessed to have Daniel's patronage and friendship 🙏 Blue Skies Forever, QED 💙
Market Design/Entrepreneurship Professor @HarvardHBS & Faculty Affiliate @Harvard Economics; Research @a16zcrypto; Editor @restatjournal; Econ @Quora; … | #QED
Trump-Xi talks have been, imo, very successful; it's hard to say who "won". But I think the narrative of miscalculation is missing one big piece. China has *tested* the export control machine. They've sampled reactions, maybe data submissions. In a year, it'll be *stronger*.
We're in a race. It's not USA vs China but humans and AGIs vs ape power centralization. @deepseek_ai stan #1, 2023–Deep Time «C’est la guerre.» ®1
ArkType 新出的 ArkRegex 是最近最惊艳的项目： https://t.co/Ffuymprv8A 正则对象 Regex 原生就是弱类型，arkregex 让它变成「强类型」，匹配/捕获直接有精确的 template literal types，结果能直接输出出来，对于学习测试正则还有好处，编译时就知道结果。 真的是很惊艳，而且测试特别完备：https://t.co/KewTTsWOxS 以后写正则必须用这个。
独立开发者 自由职业 作品 - 简单简历 https://t.co/xMu5JFIGnr 五分钟打造程序员的金牌简历 课程 - 慕课网精英讲师 https://t.co/NTyFFrvHwL 经历 - 不上班的1000天 https://t.co/bonuLQCCsY 视频 - https://t.co/aQYLgujIyC
The older I get as a bootstrapped founder the more I think of these as completely different. The outcome is similar, you build a company. But the means to get there is incomparable. If I met a VC-backed founder we would have almost no common ground, nothing to chit chat about.
To clarify, I'm not saying one is better than the other, or one requires more intelligence. I just see them as totally different skills, to the point where comparing them is almost moot.