Agentic Browsing and Copyright Infringement Issues in AI Training

A few days ago, just for fun, I checked this blog's page load speed on Google Page Speed Insights. I was intrigued by a new metric called Agentic Browsing (AI-agent-based browsing). This metric measures how user-friendly a website is for an AI agent to read and navigate. It can even autonomously click, fill out forms, and make purchases on behalf of the user. But Agentic Browsing is still under development and subject to change. In the past, search engines like Google, Bing, or Yandex only presented search results. It took time to open dozens or even hundreds of websites one by one to find the most relevant answers. A hassle, right? In the era of AI-powered personal assistants, AI agents summarize information and directly answer user's questions. They can do this because AI models are trained on vast amounts of data collected from across the web. I prefer to call it secretly scraping/collecting data without the copyright holder's permission.

This raises the question: Are AI developers violating copyright? Just out of curiosity, I searched the internet using the keywords: Are many publishers, media companies, and content creators filing lawsuits against AI developers? The results were surprising. It turns out that many media companies, book publishers, musicians, and authors have filed lawsuits. So, what do you think about this case? AI reads your work online (images, writing, designs, research, ideas, etc.), extracts the key points, and delivers it directly to paid subscription users on the AI chat page. Your work is used to power the AI. Users no longer need to visit your website, no traffics, no organic visitors, nothing.

I think the legal, ethical, and moral debates surrounding this issue will be a long and winding road. Those who support AI training, especially AI companies, argue that it falls under the concept of fair use. AI developers argue that the way AI learns is similar to how the human brain works. AI doesn't copy; it learns. The analogy is: if you read and study 100 marketing books at the library, and then write a new article about successful marketing techniques based on the insights gained from those 100 books, are you violating the author's copyright? Of course not. AI developers claim their systems do the same thing. They call it transformative work.

AI developers also argue that restrictions on AI training hinder innovation and its benefits for humanity. For example, AI can help doctors decide more quickly and accurately on necessary procedures. AI excels at analyzing medical images, such as X-rays, MRIs, and CT scans — faster and more accurate than the human eye. This reduces the rate of misdiagnosis.

Someone might ask what is my opinion on this matter? Legally, morally, and ethically, it is a violation. Even if AI doesn't copy completely, it steals ideas, patterns, and concepts — this is a serious ethical violation. Regarding fair use, I wouldn't want my work to be used to power a giant, multi-billion-dollar AI company. Everyone produces work by sacrificing time, effort, thought, money, or even love. Imagine all of that being harvested by a corporation to build paid products, such as premium subscriptions, business APIs, or to boost the company's stock valuation to billions of dollars. That is no longer fair use, but rather the unilateral exploitation of creative work. It would be "fair use" if AI were completely free for everyone to use.

Returning to Agentic Browsing, it seems the debate surrounding AI copyright is driving pressure for websites to have greater control over how AI systems access and use their content. You can decide whether to adapt to Agentic Browsing metrics or close the doors of your digital home to AI bots. You can block AI bots in your robots.txt file (search online for instructions). This step reinforces the principle: Your work is free for fellow humans who want to learn, but tightly closed to billion-dollar corporate robots. Will AI bots comply with this prohibition? Hmm… I don't know. Perhaps the simplest and most effective prevention is to keep your work under your pillow and don't upload it to the internet.

In closing, I think this is no longer just a legal and ethical debate, but extends to the broader issue of fair revenue distribution to copyright holders. Agree?