Honeypotting is an old security idea applied to a new problem: how to make large-scale scraping less attractive.
Digiday’s article explains a new defensive tactic being explored by publishers and e-commerce brands: LLM honeypotting. The idea is to lure unwanted AI crawlers into plausible but useless content, such as infinite page mazes or statistically coherent nonsense, so scraping becomes more expensive and less valuable.
For publishers, AI training on their content without permission or payment is a fundamental threat to their business model. If a scraper has to spend more processing power, time and money to collect lower-quality data, model builders will need to find more ethically sound ways to train them.
The approach is still early and imperfect. What’s clear is that AI search and LLM training are forcing content owners to rethink value, access and control. The web was built to be read. Now it is consumed at scale by machines, and with the legal system too slow to respond, content owners need other ways to fight back.
