以下文章来源于差评X.PIN ，作者江江 差评X.PIN . Debug The World，关注科技、数码、汽车、产经、游戏，传播能改变世界的科技互联网信息。 就在前不久，江江发现， 不管是点击式，九宫格选图，甚至隐式验证，AI agent 居然都能过了。 点击式验证码是最容易破防的。早在 OpenAI 的 Operator 七月刚出世的时候，就有人把它拉来对着 Cloudflare 测了一波。 本文来自微信公众号： 差评X.PIN ，作者：莫莫莫甜甜，编辑：江江、面线，题图来自：AI生成 抢票的时候，突然弹出来这么个玩意儿，你能选对吗？ 换个拼图，你每次都能一口气对准吗？ 就连有时候打个勾，你可能都得多试两次。 和验证码斗智斗勇这么多年，说实话，咱都有点麻了。 但要是我告诉你，现在上面这些验证码已经拦不住 AI 了，卡关的只有我们这些 100% 纯人类呢？ 让机器人自己点击 " 我不是机器人 "，评论区不少人都表示，这实在是有点幽默了。 还有下面这种九宫格点选，老兄自己手搓 agent，后台直连 GPT，立刻实现像素级的识别。 管你是小饼干还是小蛋糕，楼梯还是红绿灯，它看得比人都清楚。 除了显式点击，隐式 ...

Core Points - The article discusses the ongoing battle between AI and anti-AI technologies, particularly focusing on CAPTCHA systems and their evolution in response to advancements in AI [3][41]. - It highlights the challenges faced by users due to increasingly complex verification processes, which are designed to distinguish between human and automated interactions [15][43]. Group 1: CAPTCHA Evolution - CAPTCHA was originally designed to differentiate between human users and automated scripts, emerging as a response to the rise of spam and malicious activities online [4][6][8]. - Early CAPTCHAs were simple, but as OCR technology improved, more complex forms of CAPTCHA were developed, including image and audio variations [9][11]. - Modern CAPTCHAs utilize AI algorithms to create dynamic and challenging verification tasks, making it increasingly difficult for both humans and AI to solve them [12][15]. Group 2: AI Countermeasures - AI technologies have also evolved to bypass CAPTCHA systems, employing techniques such as simulating human behavior and using deep learning to recognize patterns in CAPTCHA images [13][22][30]. - Advanced AI can manipulate mouse movements and typing rhythms to mimic human interactions, making it harder for systems to detect automated scripts [22][23][30]. - AI can also utilize distributed requests and intelligent scheduling to avoid detection by anti-automation measures, maintaining a low profile while executing tasks [30]. Group 3: Future of Verification - The article suggests that the ongoing "cat and mouse" game between AI and CAPTCHA will continue, with both sides developing increasingly sophisticated methods [31][41]. - There is a growing trend towards using more user-friendly verification methods, such as behavior analysis and device reputation, to enhance user experience while maintaining security [16][20][46]. - The future may see a shift towards biometric verification methods, which, while secure, are also vulnerable to AI attacks, indicating a need for continuous evolution in security technologies [33][40].