If you've been writing online for more than a few years — blog posts, guides, forum answers, LinkedIn articles, Twitter threads — there's a reasonable chance that some of what you wrote is now living inside a large language model. It was scraped, tokenized, weighted, and baked into the parameters of a system that will answer other people's questions using a distillation of your work, without attribution, without compensation, and without asking.
This isn't a conspiracy theory. It's how these systems were built. OpenAI, Google, Anthropic, and others trained their models on the public web. Your how-to guide, your in-depth breakdown, your hard-won case study — all of it was grist for the mill. The question of whether that's legal is actively being litigated in courts across the United States. The question of whether it's fair is something every content creator and knowledge worker is quietly grappling with.
But there's a more interesting question underneath the legal one: what does this mean for the economics of sharing expertise publicly?
Chart: AI Content Growth Over Time
Needs real cited data before publishing. This chart should show the growth of AI-generated internet content (2020–present) vs. verified expert-written content.
For a long time, publishing your expertise online was a reasonable trade. You gave away content; in return you got traffic, authority, and occasionally clients. The deal made sense when Google was the intermediary — readers came to your site, you built an audience, you converted some of them into buyers. Now the intermediary is an AI, and it doesn't send traffic anywhere. It absorbs your answer and delivers it to the user directly. The trade that made public expertise valuable — visibility in exchange for knowledge — has fundamentally changed.
This is already showing up in the numbers. Organic search traffic to informational content is declining across most categories. "Zero-click" searches — where the user gets their answer without visiting a website — have become the majority of searches. For experts who built their reputation and their pipeline on content marketing, the structural underpinning of that strategy is eroding.
None of this means you should stop sharing what you know. It means the venue and the model need to change.
The most durable thing an expert can do right now is create a knowledge asset that isn't freely harvestable — one that delivers genuine value without giving that value away indiscriminately to anyone who asks a chatbot. That's a harder design problem than publishing a blog post, but it's a more defensible one. When you build a knowledge vault, you're not hiding your expertise — you're structuring it in a way that creates a real transaction between you and the people who benefit from it.
The AI training question is unlikely to resolve cleanly. Courts will issue rulings, legislation may follow, opt-out registries will be created, and none of it will fully restore the old arrangement. What will remain constant is this: the underlying expertise — the judgment, the pattern recognition, the hard-won frameworks — is still scarce. AI can summarize what's been written. It cannot replace the reasoning of someone who has actually done the thing.
The experts who figure out how to separate "my public presence" from "access to my actual knowledge" will be in a much stronger position than those who continue treating the internet as a free distribution channel for their best thinking. The former builds an audience. The latter builds a business.
If you want to see what that looks like in practice, find expert knowledge from practitioners who are already doing it. The shift is already underway.