SiteGround Hosting Blocks AI Training Bots

After completely blocking AI bot access to all sites on their shared hosting in August 2025, SiteGround relented and announced that they would allow AI bots “that are used in chat sessions”, while still blocking AI training bots. Read about it here.

What that means is that now, if you explicitly ask Chat GPT to read your site during a conversation, it will.

However it does NOT mean that Chat GPT will show your site in future generic search results where the user doesn’t specifically ask about your site! See my findings below about whether AI “learns” from searches done in chats.

Does AI Learn from Chats?

To test whether Chat GPT would incorporate what it learned in a chat into its answers to other users, I did the following experiment.

Important note: the site I used for this experiment is NOT on SiteGround hosting. I didn’t have a site on SiteGround with content that was new enough to do this experiment on (i.e., that hadn’t already been incorporated into AI training), so I used one of my personal blogs on another host where AI training bots are blocked. I think that is OK since I’m trying to answer the question of whether Chat GPT incorporates what it learns in its conversations in future answers.

Here’s what I did:

• First, I did a query on a topic that I had recently written about, and confirmed that it did NOT display my content in its answer.
• Then, I asked it to read my article, which it did, and comment on whether the results were useful and relevant, and it said they were.
• Finally, I asked it whether it would incorporate these new ideas into future query answers.

The Answer: Nope!

This is the answer Chat GPT gave:

Here’s how it works: I don’t permanently modify my general training or global behavior when I “see” a webpage like that, because I don’t retain or reuse specific external sources automatically. However, within this conversation, I can absolutely integrate the ideas from that article — and if you want, I can remember to incorporate its core insights whenever you ask me about Divi performance or CLS in the future.

So the answer is “no”, Chat GPT will not incorporate what it learns in a conversation into future answers to other people!

SiteGround’s Policies

Here’s what SiteGround says about what they are blocking:

We block crawlers intended for AI model training to protect our clients’ website data and intellectual property and save resources.

These bots attempt to collect large volumes of information to feed into training datasets, which could then be reused to build or improve AI systems without your consent.

Blocking them means your content remains under your control and is not repurposed for activities beyond its original intent.

Here’s what they say about the effect on AI query answers:

We allow crawlers from major AI providers that are used in user chat sessions.

What this means:

• Your website can be accessed and referenced in real-time answers that users receive when using these AI chat tools.
• This can increase exposure and reach, since your content may surface directly in AI chat sessions when relevant to user queries.
• The crawlers only read and display information — they should not be storing or reusing your data.

For example, imagine you run an online store hosted on SiteGround that sells handmade jewelry.

If someone asks ChatGPT “Where can I buy custom silver bracelets?”, the bot may crawl your store and display information from your product pages in its answer.

This way, potential customers could discover your website directly through the AI chat tool.

I take this to mean that if a user asks a question that causes Chat GPT to search the web, your site WILL be visible to it, but the AI won’t inherently know about your site otherwise.

I believe there are a lot of queries that Chat GPT tries to answer on its own, without crawling the web, and not having the AI trained on a commercial site is extremely detrimental to a site that you want publicized!

The Workaround

SiteGround does provide a workaround to allow AI training bots, outlined in their article:

If you have a specific reason to allow such crawlers to access your site, you have to submit an inquiry through our Help Desk > Other > AI Crawlers Setup.

Our support team will review your request and apply the change if appropriate.

They now have a nicer UI to do this:

I did this process, which required me to explain the reason for the request in a ticket. A few minutes later, I got a response letter, basically saying, “Are you sure you want to do this?”, and I had to explain myself again.

Needless to say, this is kind of a pain to do on dozens of sites! I’d much rather have a simple slide toggle to enable access to AI training bots!

What do you think? Please comment below! – Brian

Brian Shim

I am a freelance web developer and consultant based in Santa Monica, CA. I’ve been designing websites using WordPress and from scratch using HTML, CSS, PHP, and JavaScript since 2010. I create websites and web applications for businesses, nonprofits, and other organizations. I have a degree in Electrical Engineering (BSEE) from California Institute of Technology and a degree in Engineering Management (MSEM) from Stanford University. If you need help with your site, you can hire me!

brianshim.com