新数据显示，人工智慧相关的自动化抓取已成为可量化的流量来源，且呈上升趋势。某追踪抓取活动的报告估算：二○二五年第四季，其客户网站的造访中，平均每三十一次就有一次来自人工智慧抓取机器人；同年第一季则约为每二百次一次，显示季度间占比大幅提高。另一家网路基础设施公司提供的资料也指出：自二○二五年七月以来，与训练相关的机器人流量持续上升；同时，为人工智慧代理即时撷取网页内容的全球机器人活动亦在增加。
对抗手段与绕过行为同时升级。二○二五年第四季，超过百分之十三的机器人请求被估计在绕过网站用以标示抓取边界的规则档；该报告称，无视该规则的占比自二○二五年第二季至第四季增加了百分之四百。过去一年，尝试封锁此类机器人的网站数量增加了百分之三百三十六。报告并描述，部分抓取行为透过伪装成一般浏览器或模拟人类互动来降低可侦测性，导致某些人工智慧代理的流量特征几乎难以与人类区分。
生态扩张带来成本、诉讼与新商业模式并行。报告辨识出超过四十家行销可为人工智慧训练或其他目的收集网页内容的公司，显示供给端在扩张；需求端则由人工智慧搜寻与可即时取用网页资讯的工具推升。与全面封锁并行的另一条路径，是对机器存取进行计费或转向内容可见度优化，形成新的机器对机器价值交换与行销管道。文末更正指出：二○二五年第四季与人工智慧机器人占比相关的流量数字曾被修正，提示该类估算存在口径与校正风险。
New data suggests AI-related automated scraping is now a measurable—and rising—source of web traffic. One report that tracks scraping activity estimates that in Q4 2025, an average of 1 out of every 31 visits to its client sites came from AI scraping bots; in Q1 2025 it was about 1 out of every 200, implying a sharp quarter-to-quarter increase in share. Separately, an internet infrastructure company reports that training-related bot traffic has been steadily rising since July 2025, and that global bot activity fetching web pages in real time for AI agents is also increasing.
Defenses and evasions are escalating at the same time. In Q4 2025, more than 13% of bot requests were estimated to be bypassing robots.txt-style rules files that indicate scraping boundaries, and the report says the share ignoring those rules rose by 400% from Q2 2025 to Q4 2025. Over the past year, the number of websites attempting to block these bots increased by 336%. The report also describes scrapers disguising themselves as normal browsers or mimicking human interactions, making some AI-agent traffic patterns nearly indistinguishable from real users.
As the ecosystem expands, costs, lawsuits, and new business models are emerging in parallel. The report identifies more than 40 companies marketing tools to collect web content for AI training or other purposes, while demand is pushed by AI search and tools that can access live web information. Alongside outright blocking, another path is charging for machine access or shifting toward content visibility optimization (e.g., optimizing for AI-agent discovery), creating new machine-to-machine value exchange and marketing channels. A correction notes that the Q4 2025 AI-bot traffic share figure was revised, highlighting that these estimates can vary with methodology and calibration.