Your robots.txt is probably blocking the exact AI bots you need for citations — and you don't know it.
- There are two kinds of AI bots: training crawlers (GPTBot, ClaudeBot, etc.) that collect data to build AI models, and search indexers (OAI-SearchBot, Claude-SearchBot, PerplexityBot) that drive live citations in Google AI Overviews, ChatGPT, Claude, and Perplexity. Blocking training crawlers is fine, but...
- Blocking search indexers kills your citation eligibility entirely.
- Most sites that tried to protect their content in the past blocked both at the same time, by accident
- The fix: block training crawlers individually by name in your robots.txt file, leave search indexers out of the block list so they fall through to your catch-all User-agent: * rule
- robots.txt is a request, not a lock. The major AI companies honor it, but it's voluntary
- Also check: nosnippet meta tags and X-Robots-Tag headers, which can block citation eligibility even on a fully crawlable page
Your robots.txt File: One Reason You’re Not Appearing in AI Citations
You've done everything right. Structured content. Schema markup. Direct, concise answers targeting the exact questions people type into ChatGPT, Claude, and Perplexity. You've read the AEO and GEO playbooks. Your content is genuinely good.
And you're getting zero AI citations.
Here's one thing that could be causing that...
For AEO (Answer Engine Optimization) and GEO (Generative Engine Optimization) to work, AI bots need to be able to crawl your pages. If they can't access your content, they can't cite it.
The file that controls that access is your robots.txt. Unfortunately, many sites have it wrong because they block the wrong AI Bots in that file.
The problem isn't your content. The problem is you locked the door before the bots ever arrived.
How common is this? I haven't exhaustively tested yet, but I research hundreds of sites a week and typically find that 1-3 of the sites that show up in the top ten Google search results for almost any search can't be accessed by AI.
Killing their ability to be recommended by AI.
Are you one of them? I wouldn't be surprised if you are.
What Is an AI Citation?
An AI citation is when an AI tool like Google AI Overviews, ChatGPT, Perplexity, Claude, etc. answers a question and names your page as a source or recommendation.
It is not a ranking, not a position in a list. It's your site, attributed by name, linked, possibly with an excerpt pulled from your content, sitting right next to the answer the AI just gave.
Google AI Overviews pull sources into a collapsed block around organic results. ChatGPT surfaces a source panel. Claude links inline when it's pulling from the web. Perplexity shows citations on every single response. It's built that way by design.
Most people still think of search as a list of 10 blue links. You rank, someone clicks. That model still works at the moment (but Google's changing it as we speak.)
But the visitor who clicks a citation is a different person from the one who clicks the fourth result in the list. Citation clickers have already built some trust from the answer AI's response. They're clicking to learn more from the source the AI pointed them to. That's a warmer, higher-intent visit than almost anything organic search delivers.
And here's the thing most sites are missing right now: citation competition is still wide open. Most sites haven't optimized for it at all. The ones that figure this out, getting access and content right, are filling a gap that won't stay open long.
Access comes first. Everything else follows.
Why Your robots.txt Matters for AIO, AEO, and GEO
AIO, GEO, and AEO all depend on one thing: the AI bot being able to read your page.
For example, before a page is eligible for AI citation in Google AI Overviews, it has to be a regular indexable page. Block the AI bots, and you block the citation. No amount of schema markup overrides that.
The same applies to GEO; AI generally has to see the page to recommend it.
Does this sound familiar? You've done all the optimization work, and nothing worked.
Check the door first. If you're blocking AI from seeing the pages on your site, it's unlikely to show you in its results.
You Also Need to Allow Snippets
There's a second layer to check. Google requires that a page allow it to pull snippets before it will include it in AI results.
So, even when your robots.txt looks clean, if you've added nosnippet meta tags or max-snippet:0 HTTP into your headers, you will block AI answer eligibility at the page level. Be sure to check your page meta language templates, not just your robots.txt.AI Bots to Allow in Your robots.txt file
The search and retrieval bots you generally want to allow:
OAI-SearchBot: supports indexing and retrieval for ChatGPT Search and citations in ChatGPT responses
Claude-SearchBot: supports indexing and retrieval for Claude web search features and citations
PerplexityBot: supports indexing and retrieval for Perplexity answers
Googlebot: required for traditional Google Search visibility and strongly tied to eligibility for AI Overviews and other Google AI surfaces
And then there are the retrieval bots
In addition to search indexers, some AI platforms use separate user-initiated retrieval agents that fetch pages in real time when a user asks a question. These are distinct from training crawlers.
The retrieval bots you generally want to allow:
ChatGPT-User: OpenAI’s user-initiated retrieval agent for live webpage fetching inside ChatGPT
Claude-User: Anthropic’s user-initiated retrieval agent for live webpage access in Claude
Perplexity-User: Perplexity’s user-initiated retrieval agent for live answer generation
Blocking these retrieval agents may reduce your likelihood of being cited or retrieved in AI-generated answers, even if your pages remain indexed.
What You Want to Block in Your robots.txt File
Training crawlers build model weights, not citations. These bots collect web content to train future AI models. Blocking them generally opts your site out of training-data collection without materially affecting citation eligibility, as long as the search and retrieval bots above can still access your pages.
The training crawlers you can often block with minimal impact on AI citation visibility:
GPTBot: OpenAI training data collection
ClaudeBot: Anthropic training data collection
Google-Extended: Google Gemini and Vertex AI training controls
Applebot-Extended: Apple Intelligence training controls
Meta-ExternalAgent: Meta AI training
CCBot: Common Crawl dataset collection used by many AI pipelines
Bytespider: ByteDance/TikTok crawling and training
cohere-ai: Cohere training crawler
Amazonbot: Amazon crawling and AI-related indexing/training
Diffbot: large-scale data extraction and knowledge graph creation
AI2Bot: Allen Institute for AI research datasets
One important caveat: crawler behavior and robots.txt enforcement policies continue to evolve. Some companies clearly state that their retrieval agents honor robots.txt, while others provide more limited guarantees. Website owners who require strict access control should not rely solely on robots.txt and may also need authentication, WAF rules, CDN protections, or IP-based controls.
The AI-Blocking Problem Started Years Ago
Here's what happened in 2023 and 2024. Site owners like you and me saw headlines about AI companies scraping the web, got scared that all our hard work was going to be added to the Borg, and instituted blanket AI bot blocks.
Reasonable instinct. Bad execution.
Site after site blocked training crawlers AND search indexers at the same time. We thought we were protecting our content. We were also unknowingly opting out of every AI citation.
We had no idea.
And since we think we're "protected from AI stealing our content," many site owners have never checked whether that assumption is correct.
The key to blocking the AI Bots we don’t want while allowing the citation and retrieval bots free access is your robots.txt file.
Configuring Your robots.txt to Allow the AI Bots You Want, Blocking Those You Don't
Your robots.txt file is a plain-text file at the root of your website that tells web crawlers which pages or sections of your site they can access.
Let’s start with an optimized example of what a good one could look like.
This is the actual configuration I have running on doncrowther.com that I recommend you use too. This is not a template. Not a hypothetical. Here's what I’m really using. (I'll explain why each of these elements is included below):
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73
# ==============================================
# Block AI training crawlers
# ==============================================
User-agent: GPTBot
Disallow: /
User-agent: ClaudeBot
Disallow: /
User-agent: Google-Extended
Disallow: /
User-agent: Applebot-Extended
Disallow: /
User-agent: Meta-ExternalAgent
Disallow: /
User-agent: CCBot
Disallow: /
User-agent: Bytespider
Disallow: /
User-agent: cohere-ai
Disallow: /
User-agent: Amazonbot
Disallow: /
User-agent: Diffbot
Disallow: /
User-agent: AI2Bot
Disallow: /
# ==============================================
# All standard search crawlers
# ==============================================
User-agent: *
# Admin/login
Disallow: /wp-admin/
Disallow: /wp-login.php
Allow: /wp-admin/admin-ajax.php
# WordPress/system
Disallow: /xmlrpc.php
Disallow: /readme.html
Disallow: /license.txt
# API/privacy
Disallow: /wp-json/wp/v2/users/
# Duplicate/low-value pages
Disallow: /?s=
Disallow: /search/
Disallow: /?p=
# Pagination duplicates
Disallow: /tag/*/page/
Disallow: /category/*/page/
# Cache/plugin artifacts
Disallow: /wp-content/uploads/wpo-plugins-tables-list.json
# Allow rendering assets
Allow: /*.css$
Allow: /*.js$
Sitemap: https://doncrowther.com/sitemap_index.xml
Why this robots.txt is Structured the Way it is
Every line in this robots.txt file serves a specific purpose. Here's why they're there:
Section 1: Training crawler blocks
This section contains eleven explicit blocks. Each bot gets its own stanza. This is not redundant formatting. It's necessary to block each of them individually.
If you used User-agent: * to block everything, that would catch the search indexers too: the exact bots you need to allow. This process of surgical naming is the entire point.
Disallow: / blocks those bots from the whole site. Not a folder. Not a section. Every page, every URL, nothing gets through for that crawler.
Section 2: User-agent: * (opening things up for the search/citation indexers we want)
Because the AI bots we want to access the site, like OAI-SearchBot, Claude-SearchBot, and PerplexityBot, aren't named in any block above, they hit this catch-all rule. They’re allowed to read everything except the paths you see I've explicitly blocked here. That's how they end up with access to your content.
Admin and login paths
/wp-admin/ is your site's backend: your settings, your users, your drafts. There's nothing a bot needs to see in those files.
/wp-login.php is the login page. Blocked. You don't want that page showing up in the search engines.
The Allow: /wp-admin/admin-ajax.php line is a WordPress quirk you need to know about. Some front-end functionality, like dynamic content, live search, and contact forms, routes through that endpoint. Block it, and things break for real users. Allowing access to it is deliberate.
WordPress system files
xmlrpc.php is an older API endpoint and a known attack vector. There is no legitimate indexing reason to allow it.
readme.html and license.txt expose your WordPress version number. They have no SEO value. They're also a minor security risk. This is an easy block.
API and privacy
/wp-json/wp/v2/users/ is a WordPress REST API endpoint that returns a list of your site's usernames. It's public by default for some reason. It shouldn't be. Block it.
Duplicate and low-value pages
?s= and /search/ are your site's internal search result pages. They generate unique URLs for every search query, but the content is thin and inconsistent. That's crawl budget waste at scale.
?p= is the numeric post ID format WordPress generates alongside your clean slug URLs. You don't need both indexed. Block the ID format. (Of course, you want to be using page titles as your URLs. If you're not, you'll want to allow this parameter.)
Pagination duplicates
Category and tag archive pages past page 1 are repetitive and thin. Page 1 of a category archive has value. Page 47 of /category/marketing/ does not. Keep page 1. Block the rest.
Cache and plugin artifacts
The WPO plugins JSON file is internal housekeeping for a caching plugin. There's no content there. No one should be indexing it.
Rendering assets
This one matters more than most people think. The Allow rules for CSS and JS files are not optional if you care about how Google reads your pages. (And yes, for you purists out there, I realize there's redundancy here...)
Googlebot renders pages before making quality judgments. It's not reading raw HTML like it's 2005. Block your CSS and JS files, and Googlebot sees unstyled markup. It misses layout signals, structured elements, and visual context that factor into quality assessment. These allow rules are cheap insurance.
Sitemap
The sitemap line tells every crawler that can access your site, including the search indexers you've allowed, exactly where to find your content. Don't skip it. Don't link to a broken URL. Verify that your equivalent of https://doncrowther.com/sitemap_index.xml resolves before you push this file.robots.txt Is a Request, Not a Lock
Here's something most guides leave out, but you should know:
robots.txt is a voluntary standard. There's no technical mechanism that prevents a bot from ignoring it. You publish the file, you signal your preferences, and then you trust that the bot's operator chose to build a compliant crawler.
Malicious scrapers, spammers, and shady data resellers? They ignore it regularly. That's a separate problem robots.txt was never designed to solve.
But here's why this doesn't undermine what you just set up.
The bots in this configuration: GPTBot, ClaudeBot, OAI-SearchBot, PerplexityBot, Google-Extended, are run by companies with legal teams, public commitments, and business incentives to behave. OpenAI has explicitly documented that GPTBot honors robots.txt. Anthropic has done the same for ClaudeBot and Claude-SearchBot. Google has done it for Googlebot, Google-Extended, and the rest of its fleet.
These companies are probably not going to blow up their legal exposure, their publisher relationships, and their public trust to grab your content after you've told them not to. The risk-reward doesn't work for them. Compliance is in their interest.
So for the bots that matter to AEO and GEO, robots.txt works. Not because it's technically enforced. Because the companies running these bots have chosen to make it work.
That said, if you want a hard technical block rather than a polite request, one that issues a 403 before the bot even reads your file, that's what Cloudflare's AI Crawl Control does. Robots.txt and Cloudflare serve different purposes. This article covers one. Another article will cover Cloudeflare.
llms.txt: The Emerging Complement to robots.txt for AI Access
robots.txt controls access. llms.txt is a proposed standard for giving AI systems a plain-language summary of what your site is, what you publish, and what's worth reading.
The format is simple: a plain text file at the root of your site, written in Markdown, that describes your site's purpose and links to your most important content. Think of it as a briefing document for AI systems that have permission to read your site but don't know where to start.
It's not required. It's not widely adopted yet. Google both claims it's totally unnecessary and looks for it in certain cases. You won't get penalized for having it, and no AI system is penalizing you for not having one today.
Create your llms.txt file now. The effort is low, and the standard is forming. A few hours of work now puts you ahead of a requirement that may become expected. When it matters, you'll already have it.
Three Things to Audit Right Now
1. Check your robots.txt against the bot list above.
Open your current robots.txt and look for GoogleBot, OAI-SearchBot, Claude-SearchBot, and PerplexityBot. If any of them appear in a Disallow rule, or if you have a blanket block that would catch them, fix it. Leaving them blocked is cutting off your citation pipeline at the source.
2. Search your page templates for nosnippet, max-snippet:0, and X-Robots-Tag headers.
robots.txt is the first gate. These meta tags and X-Robots-Tag HTTP headers are the second.
A page can be perfectly crawlable but still ineligible for AI answer features because of a snippet restriction baked into the template.
nosnippet tells search engines not to show a text snippet from the page. max-snippet:0 sets the snippet length to zero. Either one in a meta tag or an X-Robots-Tag HTTP header will block your content from AI answers, even if the page is fully crawlable.
Check every page type: posts, pages, category archives, and landing pages.
3. Make sure your sitemap is declared in robots.txt and resolves correctly.
Open the sitemap URL in a browser. Does it load? Does it contain your actual content pages? A declared sitemap that 404s is worse than no declaration.
Access First. Everything Else Follows.
The reason much of the work people are doing in AEO and GEO doesn't produce results isn't bad advice. It's that this prerequisite was never checked.
You can write the cleanest, most citation-worthy content on your topic and get nothing. Nothing. Not. A. Single. Citation if the bots responsible for those citations were blocked before they ever read a word of it.
Copy the configuration above, check it against your current file, and deploy it. Then worry about what you're publishing.
What do you think? Have you checked your robots.txt lately? Let me know in the comments. And let me know if there are any improvements I should make to that document. I’m always open to feedback.
This is Don Crowther saying, just go do this stuff.
Frequently Asked Questions About AEO, GEO, AIO and AI Bots
AEO, GEO, and robots.txt basics
What is GEO optimization?
GEO (Answer Engine Optimization) means structuring your content so AI-powered search tools like Google AI Overviews, ChatGPT, Claude, Perplexity, and others surface your pages as cited sources. Instead of ranking in a list of ten blue links, the goal is to be the answer, or the source cited alongside it. It starts with making sure AI bots can actually reach your content.
What's the difference between AEO and GEO?
AEO focuses on coming up in special elements of search, like snippets, people also ask, etc. GEO is broader, covering any generative AI system that pulls from web content. In practice, the tactics overlap almost entirely. The biggest difference is that AEO tends to target traditional search features, while GEO includes Google AI Overviews and standalone tools like Claude and Perplexity as primary targets.
So you know, according to Google, it's all SEO and none of these distinctions matter.
Why does robots.txt affect AEO and GEO?
Because robots.txt is the first thing a crawler checks before deciding whether to read your site. If your robots.txt tells a bot to stay out, it stays out. No page on your site becomes eligible for citations if the bot responsible for those citations can't access the content. Every AEO and GEO tactic you apply is irrelevant until this prerequisite is met.
What happens if AI bots can't crawl my site?
They don't cite it. The AI systems powering ChatGPT citations, Claude citations, Perplexity answers, and Google AI Overviews all depend on bots that have crawled the relevant pages. A blocked bot means a missing source. Your content can be perfect: direct answers, clean structure, strong schema. It still won't appear in a single AI-generated response if the bot couldn't read it.
What is the difference between training crawlers and search indexers?
Training crawlers collect web content to build or improve AI models. Blocking them opts your site out of training data. Search indexers crawl your pages to serve live citations in AI products. They're what drives real-time answers in ChatGPT, Claude, and Perplexity. Blocking a training crawler has no effect on citation eligibility. Blocking a search indexer removes you from the citation pool entirely. Most sites that accidentally opted out of AI citations blocked both at the same time.
Should I block all AI bots to protect my content?
No. And this is the mistake that's costing sites citations right now. Blocking all AI bots simultaneously shuts out training crawlers AND the search indexers that drive live citations in ChatGPT, Claude, and Perplexity. The correct approach is surgical: block the training crawlers by name, leave the search indexers out of the block list, and let them fall through to the catch-all rule. Protecting your content from training and staying eligible for citations are not mutually exclusive. You just need the right configuration.
How do I know if my site is currently blocking AI search indexers?
Open your robots.txt file at yourdomain.com/robots.txt and search for OAI-SearchBot, Claude-SearchBot, and PerplexityBot. If any of them appear in a Disallow rule, you're blocking them. Also check whether you have a blanket User-agent: * block, then look at whether search indexers get an explicit Allow override. If they don't, they're blocked. Fix it before doing anything else for AEO or GEO.
Do AI bots actually have to follow my robots.txt?
No. robots.txt is a voluntary standard. There's no technical mechanism that forces compliance. A bot can ignore it entirely. The reason it works for the bots in this article: GPTBot, ClaudeBot, OAI-SearchBot, and the rest, is that the companies running them have publicly committed to honoring it and have business, legal, and reputational reasons to follow through. Malicious scrapers are a different story. robots.txt was never designed to stop them. If you want a hard technical block that actually denies access at the network level, that's a Cloudflare conversation, not a robots.txt conversation.
Search Indexers (allowed by default in the sample robots.txt file shown above)
What is OAI-SearchBot?
OAI-SearchBot is OpenAI's search indexer, the bot that crawls pages to serve citations in ChatGPT responses. It's not the same as GPTBot. GPTBot builds training data for ChatGPT. OAI-SearchBot handles live citations. Allowing it, by not explicitly blocking it, is how your content becomes citable in ChatGPT. If you blocked all OpenAI bots in the past without distinguishing between them, you blocked this one too.
What is Claude-SearchBot?
Claude-SearchBot is Anthropic's search indexer, separate from ClaudeBot. ClaudeBot is the training crawler: blocking it opts you out of your data being used to train Claude. Claude-SearchBot drives citations in Claude's responses. Not blocking it means Claude can cite your content. The names are similar enough that people confuse them, which is exactly how the blanket-block mistake happens.
What is PerplexityBot?
PerplexityBot is Perplexity's crawler. Perplexity is citation-first by design. Every answer it generates shows the sources it pulled from, with links back to the original pages. Allowing PerplexityBot means you're in the pool of potential sources. Blocking it means you're not, regardless of how well your content answers the question.
What is Googlebot, and how does it relate to AI?
Googlebot is Google's primary search crawler. A page has to be indexable by Googlebot before it's eligible for Google AI Overviews. If Googlebot can't reach it, AI Overviews can't surface it. Googlebot access is the baseline requirement for everything Google AI does with your content.
What is ChatGPT-User?
ChatGPT-User is OpenAI's user-initiated fetcher. When a ChatGPT user asks a question that requires pulling a live page, ChatGPT-User retrieves it in the moment. It's distinct from GPTBot, which builds training data, and from OAI-SearchBot, which maintains the search index. Blocking it can stop your page from being pulled into a live ChatGPT answer. OpenAI's crawler documentation no longer treats robots.txt as a dependable control mechanism for ChatGPT-User, so website owners should not rely solely on robots.txt if they need to restrict access. Blocking or otherwise restricting ChatGPT-User may reduce your chances of being used as a live source in ChatGPT responses.
What is Claude-User?
Claude-User is Anthropic's user-initiated retrieval agent. When a Claude user asks a question that requires accessing a webpage, Claude-User retrieves the content in real time. It is separate from ClaudeBot, which gathers training data, and Claude-SearchBot, which indexes content for search features. Anthropic states that Claude-User honors robots.txt directives, so blocking it prevents your content from being retrieved for live user queries. Leaving it unblocked allows your content to remain eligible for citation and retrieval within Claude.
What is Perplexity-User?
Perplexity-User is Perplexity's user-initiated retrieval agent. When a Perplexity user asks a question that requires live web access, Perplexity-User retrieves content on demand. It is separate from PerplexityBot, which is used for indexing. Perplexity states that its crawlers respect robots.txt directives, although its crawler practices have been the subject of public disputes and criticism from Cloudflare and some publishers. For publishers seeking visibility in Perplexity answers, Perplexity-User should generally be allowed access.
Training Crawlers (blocked in the sample robots.txt file shown above)
What is GPTBot?
GPTBot is OpenAI's training crawler. It crawls the web to collect data for training future versions of GPT models. Blocking it with Disallow: / opts your site out of that data collection. It has no role in ChatGPT's live citation system. That's OAI-SearchBot's job. Blocking GPTBot does not prevent ChatGPT from citing you.
What is ClaudeBot?
ClaudeBot is Anthropic's training crawler. It collects content for training Anthropic's models. Blocking it opts you out of training data. Claude's live citation system uses Claude-SearchBot, a separate bot with a separate user-agent string. Blocking ClaudeBot doesn't affect whether Claude cites your content.
What is Google-Extended?
Google-Extended is Google's dedicated opt-out for Gemini and Vertex AI training. Blocking it prevents your content from being used to train Google's AI models while keeping Googlebot access intact for regular search and AI Overviews. Google built this specifically so sites could opt out of AI training without losing search visibility. Use it.
What is Applebot-Extended?
Applebot-Extended is Apple's training crawler for Apple Intelligence, the AI features built into iOS and macOS. Blocking it opts out of Apple AI training. Standard Applebot, which handles Siri search results and Spotlight suggestions, is a separate bot and is unaffected. The "-Extended" suffix is Apple's equivalent of Google-Extended: a surgical opt-out without broader crawl consequences.
What is Meta-ExternalAgent?
Meta-ExternalAgent is Meta's training crawler. It collects web content to train the AI models behind Facebook, Instagram, WhatsApp, and Meta AI. Blocking it prevents your content from becoming Meta training data. Meta doesn't have a public citation-focused search indexer the way OpenAI and Anthropic do, so there's no citation tradeoff. Block it.
What is CCBot?
CCBot is the Common Crawl bot. Common Crawl is a nonprofit that publishes open web archives used as training data by a wide range of AI companies. It's one of the most widely used training data sources in the AI industry. Blocking CCBot reduces the availability of your content across a broad set of AI training pipelines, not just one company's product.
What is Bytespider?
Bytespider is ByteDance's crawler. ByteDance runs TikTok. Bytespider collects data for training AI models across ByteDance's products. It's not a search indexer. There's no TikTok citation system that requires access to your content. Blocking it is straightforward.
What is cohere-ai?
cohere-ai is Cohere's training crawler. Cohere builds AI models for enterprise applications. This bot collects web content for training those models. Cohere doesn't operate a public AI answer product that cites web sources, so there's no citation consideration when blocking it.
What is Amazonbot?
Amazonbot is Amazon's web crawler, primarily used for Amazon's AI and machine learning work. Some sources suggest allowing it for Alexa voice result eligibility, though Amazon hasn't published clear documentation on this relationship. If Alexa citations matter to your site, test before blocking. If they don't, blocking is the conservative choice.
What is Diffbot?
Diffbot is a data extraction company that turns web pages into structured data and sells that data to businesses, developers, and AI companies. It's not a search indexer. Blocking it prevents your content from being packaged and resold as training or enrichment data without your knowledge or consent.
What is AI2Bot?
AI2Bot is the Allen Institute for AI's crawler. AI2 is an academic research organization that builds datasets for AI research, including the Dolma dataset used to train the OLMo family of open-source models. AI2Bot collects web content for those research datasets. AI2 doesn't operate a public AI answer product that would cite your content in user-facing responses.


