Wow, that's a great point... I hadn't considered that. I assumed it's all training.
From what I understand, Cloudflare is trying to create a way for agents to consume content in a more structured manner than allowed for attention to the author, and potentially payment along with it.
I don't want to be paid but I'd love to see how often context from my writing winds up in a session a human is actively using.
It is good to make a proper distinction, in the ChatGPT context, between crawlers and agents. The crawlers go for the content to build a new model, the agents serve content to users. The last one can be very useful.
They use different user-agent strings. The crawlers obfuscate themselves and use residential proxies. The agents call themselves ChatGPT-User. Of course Cloudflare wants OpenAI to pay them for not blocking ChatGPT-User by default.
Thanks. It seems to be very local/incidental. The page works from the locations I can test, but I’ll check whether one edge cache or request path served a bad response.
Built a Cloudflare Worker that classifies all traffic by visitor type: humans, AI crawlers, SEO bots, residential proxies. On most days, humans are a minority. The article explains what the patterns reveal and why edge logs show things your analytics tool can't.