Quick answer
An AI bot is software that uses AI to talk with people or carry out a task automatically. Owners meet the word in three situations: a bot they buy, which sellers call a chatbot, voice bot, assistant or agent; crawlers that visit their website; and spam accounts. Ask what the software does without being asked.
Key takeaways
- An AI bot is software that uses AI to talk with people or do a task automatically; owners meet it as a bot they buy, as crawlers on their website and as spam accounts.
- The word is not used consistently: Google Cloud's comparison table treats a bot as the least autonomous of three, while Amazon Web Services uses chatbot for a range that starts at rule-based.
- Ask one question of any product called a bot: what does it do without being asked? The answer separates a chat window from something that acts.
- The second situation is traffic to your site. OpenAI, Anthropic and Google each document separate controls for AI training, for search and for requests a person triggers, and their pages treat robots.txt differently for that last group.
- Blocking a training crawler and blocking a search crawler are different choices; each company's page says what the search one affects.
What is an AI bot?
An AI bot is software that uses AI to talk with people or carry out a task automatically. "Bot" is a loose word, so a seller can use it for a chatbot, a voice bot, an assistant or an agent, and your server logs use it for automated visitors. Owners meet it in 3 situations: a bot you buy, crawlers visiting your site and spam accounts. This page sorts them, and the first comes in four product words.
Which words do vendors use for a bot?
This site sorts the products sold as bots under four words: chatbot, voice bot, assistant and agent. The table is this site's glossary, built from the vendor definitions quoted below it; it is not a standard.
| Word | What it does | Who starts it | Example of the type |
|---|---|---|---|
| Chatbot | Answers questions in a chat window or by text | The visitor writes first | A website chat that answers hours and booking questions |
| Voice bot | The same, but by phone | The caller dials in | A phone line that answers, takes a message and sends it to you |
| Assistant | Helps you inside an app and suggests next steps | You ask | A writing helper that drafts a reply you then approve |
| Agent | Takes a goal and chooses its own steps | You set the goal, then it works | Software that chases an unpaid invoice across email and your calendar |
Amazon Web Services defines a chatbot this way: "A chatbot is a program or application that users can converse with using voice or text." It also writes that "A voicebot, or voice assistant, is a chatbot that listens to voice commands," so a voice bot is a chatbot you talk to.
For assistants, Google Cloud's comparison says an assistant "can recommend actions but decision-making is done by the user." The agent row is only a pointer; the live agents page covers which jobs are sold as hands-off and what a first one costs.
Does every vendor mean the same thing by "bot"?
No, and the difference can change what you buy. Google Cloud's comparison table describes a bot as "Follows pre-defined rules; limited learning; basic interactions" and adds "Bots are the least autonomous, typically following pre-programmed rules." In that table, bot is the weakest of three.
Amazon Web Services writes about chatbots rather than bots, and uses that word for a family. It says "Rules-based chatbot technology is the simplest version of chatbot software," and its page goes on to describe AI-powered chatbots as well. Google Cloud's table places the bot at the scripted end of its scale; AWS's page uses chatbot for a range that starts there. These are two vendors using nearby words differently, and neither is wrong.
So "AI bot" on a sales page tells you little. Ask whether the product follows a script you could read or generates its own answers, and get that in writing.
How do you tell which kind you are being sold?
Ask what it does without being asked. A chat window that waits for a visitor is a chatbot. Software that acts on a trigger you set, the same way every time, is an automation. Work through these checks:
- Who starts it: a visitor, you, a schedule or nobody?
- What does it do between requests? "Nothing" points to a chatbot or assistant.
- Can it change anything, such as send a message, book a slot or move money, or can it only answer?
- Where does it hand over to a person? Amazon Web Services describes chatbots that "attempt to answer your queries before passing you on to a human representative." Ask where that point is.
- Can you see what it did afterwards, in a log or a summary?
If answers one to three say "visitor, nothing, answer only", you are looking at a chatbot, and adding one to your website without coding is the relevant guide. If the product acts without a prompt, read the practical guide to AI automation first.
What does "AI bot" mean in your website traffic?
It means automated software that visits your site, and OpenAI, Anthropic and Google each document separate controls for it. A "crawler" is a program that reads web pages. Each company's page covers a training control, a search crawler and requests a person triggers. The table repeats what each company's own page says, as read on October 7, 2026; names and rules can change, so check the vendor page before you act on them.
| Company | Training | Search | A person asks | What the page says about robots.txt |
|---|---|---|---|---|
| OpenAI | GPTBot | OAI-SearchBot | ChatGPT-User | Disallowing GPTBot indicates content should not be used for training; for ChatGPT-User, rules "may not apply" |
| Anthropic | ClaudeBot | Claude-SearchBot | Claude-User | The page says, of all the bots it describes, that they respect "do not crawl" signals by honoring robots.txt directives; the Claude-User entry itself says only that site owners can control those requests and does not mention robots.txt |
| Google-Extended (a robots.txt token, not a separate crawler) | Googlebot | User-triggered fetchers, for example Google-Agent | Common crawlers "always obey robots.txt rules when crawling automatically"; user-triggered fetchers "generally ignore robots.txt rules" |
These three statements are not about identical requests. OpenAI and Anthropic both describe an assistant visiting a page when a user asks a question. Google's list is broader: its user-triggered fetchers include feed readers, link previews and Google-Agent, which Google says is "used by agents hosted on Google infrastructure to navigate the web and perform actions upon user request."
Read the last column as what each page states, not as a ranking. All three publish IP ranges you can check your logs against. OpenAI's page lists "Published IP addresses" for GPTBot, OAI-SearchBot and ChatGPT-User, Anthropic's says that a crawler with a source IP address on its list "indicates that the crawler is coming from Anthropic," and Google's says the IP ranges its user-triggered fetchers use are published as JSON files.
Google adds "The user agent string can be spoofed." Anthropic cautions about blocking by IP: "Alternate methods like blocking IP address(es) from which Anthropic Bots operates may not work correctly or persistently guarantee an opt-out, as doing so impedes our ability to read your robots.txt file." A spoofing or IP caution is not stated on the OpenAI page we read.
Spam bots that post to forms or comments are not covered by any of these pages. This site's view: use the spam filter in your form tool or hosting plan, not robots.txt.
Should you block AI crawlers?
This site's view: treat the training switch and the search switch as two separate decisions. OpenAI's page says its settings are independent of each other, and Anthropic and Google list training and search as separate controls. Each company says what the search side affects:
- OpenAI: "Sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though can still appear as navigational links."
- Anthropic: disabling Claude-SearchBot "may reduce your site's visibility and accuracy in user search results." It says the same of disabling Claude-User, for "user-directed web search."
- Google: "Google-Extended does not impact a site's inclusion in Google Search." Preferences set on Googlebot, by contrast, "affect Google Search."
Googlebot is Google's general search crawler, so Google's search choice is not an AI-only switch.
This site's view: if you want to appear in ChatGPT's or Claude's answers, blocking their search crawlers may remove you, while blocking the training crawlers is a separate choice.
The controls live in a robots.txt file at the top of your website. Crawler names and rules are as each vendor's page showed them on October 7, 2026; if a name has changed, the vendor page linked in the sources is the one to follow. If you use a builder or an agency, ask them to show you the file.
What should you do next?
Pick the situation that applies to you today.
- Buying a bot: ask the seller which of the four words it is and what it does without a person.
- Crawlers: decide training and search separately, and check your robots.txt against the vendor pages on the date you read them.
- Spam: use your form tool's or host's spam filter.
How was this page made?
The vendor and documentation pages listed in the sources were read on October 7, 2026, from where this page was read, and the quotations come from them. This page was drafted with Claude, an AI model made by Anthropic, and checked against each vendor's published material. Anthropic is one of the three companies covered in the crawler table and was described only from its own help page, in the same way as OpenAI and Google. It recommends no product.
FAQ
Is a bot the same as a chatbot?
No. A chatbot is one kind of bot: Amazon Web Services defines it as a program people can converse with by voice or text. A bot can also be a crawler that reads web pages or a program that runs a task with no conversation at all.
Is an AI bot the same as an AI agent?
Not always: sellers call both bots, and the live agents page covers what an agent is and which jobs to hand over.
Are AI bots on my website good or bad?
Both exist. The crawlers that OpenAI, Anthropic and Google document are published by named companies, and their pages describe robots.txt controls and published IP ranges you can check logs against. Spam bots that post junk on forms and comments are a separate problem that these pages do not cover.
Can I stop an AI bot from reading my website?
Each of the three companies documents a robots.txt control for its crawlers. For requests a person triggers the pages differ: OpenAI says robots.txt rules may not apply to ChatGPT-User, Google says its user-triggered fetchers generally ignore robots.txt, and Anthropic says Claude-User lets site owners control those requests. They are not all the same kind of request, so read each page before you rely on it.
Does blocking an AI crawler remove me from Google search?
Google says its Google-Extended token does not affect inclusion in Google Search. OpenAI says sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as navigational links, and Anthropic says disabling Claude-SearchBot may reduce a site's visibility in its search results.