Every answer engine has a crawler, and every crawler reads robots.txt. Blocking them all is a decision some publishers make deliberately. Blocking them by accident, because a plugin or a host added a rule years ago, is far more common.
Which bots are we talking about?
OpenAI runs GPTBot for training and OAI-SearchBot for ChatGPT search results. Anthropic runs ClaudeBot. Perplexity runs PerplexityBot. Google-Extended controls whether your content feeds Gemini, separately from ordinary Google search. Bingbot feeds both Bing and Copilot. Then there are CCBot, Applebot-Extended, Bytespider and Meta-ExternalAgent, each with their own purpose.
Search yes, training no
The most defensible policy for a business site is to allow the bots that send visitors and to decide separately about the ones that only train models. That means allowing OAI-SearchBot, PerplexityBot and Bingbot, and making a conscious choice about GPTBot, ClaudeBot, Google-Extended and CCBot. Rankavia ships this as a one-click preset, writes the rules into robots.txt, and shows you what each bot actually fetched in the AI bot log.
How to check what you have today
Open /robots.txt on your own domain and look for any User-agent block that names one of the bots above, or a blanket Disallow: /. Then check the response headers of a page for X-Robots-Tag. The free AI crawler check on rankavia.com does both and tells you, bot by bot, whether you are allowing, blocking or saying nothing.
Saying nothing is allowing. If that is not what you intend, write it down.