I feel like most casual users would not make the connection of "crawlers" to link previews that they talk about it the article.

Sure, if you understand that robots.txt includes all robots then sure. But that is not how general news media has been talking about robots.txt.

Ok. That quotes a tweet by Cloudflare's CEO. IDK what his qualifications are, but his conflict of interest is obvious enough. Real quality journalism there.

ETA: I looked at what the Cloudflare CEO said again. To be fair to him, he is not actually claiming that Googlebot collects AI training data. He's talking about the AI overview, which is a search feature. The data for search features is collected by Googlebot. I'm not sure why someone would want their link listed in search but not appear much more prominently in the AI overview.

Here's Google technical documentation on its crawlers: https://developers.google.com/search/docs/crawling-indexing/google-common-crawlers

that is not how general news media has been talking about robots.txt.

Ahh, yes. I think there is a lesson there.

So what's the quote from the documentation that backs up your claim? The line "perform other product specific crawls" seems extremely vague by design.

I'm not really sure what you are asking here. Did you notice that you can scroll down and see a list of their crawlers?

Nothing on this page seems to contradict the article. But if I simply missed the part that does, I'd be happy to learn.

You look up what Googlebot does. No AI.

You want to know what crawlers do AI? Just search for "AI", or "training", or some such, or skim through. It's not long. Google-Extended collects training data. Note that Google-Extended is explicitly not used to rank pages.

Did that help?

You look up what Googlebot does. No AI.

The page seems written to perhaps suggest it but doesn't explicitly say the other bots can't feed into some other sort of AI training. It would be in Google's interest to mislead the users here.

Edit: I found a quote where it says Googlebot does both in one: "Google-Extended doesn't have a separate HTTP request user agent string. Crawling is done with existing Google user agent [...]" and I guess Cloudflare doesn't trust Google to abide by the access controls. That seems sensible to me. Edit 2: What exactly the CEO believes was perhaps rightfully disputed below, it was just my guess.

It would be a lot to write, if you had to say what something does not do rather than what it does.

I looked at what the Cloudflare CEO said again. To be fair to him, he is not actually backing you up. He's saying that Google makes no difference between the AI overview and the other search results. That is true. The AI overview is a search feature. I'm not sure why someone would want their link listed in search but not appear much more prominently in the AI overview.

But the article later does back it up: "Although Cloudflare singled out Google, other search engines that view AI search features as part of their search products also use the same bots for training as they do for search indexing."

In any case, I'm okay with admitting neither you nor me can look inside Google to see they're doing. But the claims are out there, I didn't make them up, whether they're true or not. Thank you for the certainly interesting Google crawler info link.

Kinda, but also not really. Any major tech player that has billions to lose will make a show of respecting robots.txt when presenting that information to third parties, lest they be exposed by basic journalism.

However, they also have separate networks in R&D that sweep the net all the time and do not care about such restrictions. It's theatre.

And they're still happy to punish people that have the gall to publicly decline their crawlers. Basically they can eat their cake and have it too.

But the article later does back it up

The CEO of Cloudflare did not assert that. I was surprised that he would claim such a thing, and that should have made me read more carefully. Elon Musk notwithstanding, neither incompetence nor conspiracy theorizing are common at that level, publicly anyway.

You can believe whatever you like, of course. Freedom of opinion is nothing if not the right to be wrong.

Right, but the article does. Anyway, I'm moving on. Thanks for the discussion.