Every domain in our database is classified into one of 18 functional categories and a specific subcategory — built for firewalls, DNS resolvers, and security teams. The taxonomy updates continuously as new AI tool types emerge.
Most AI directories organize by popularity or pricing — useless for security teams deciding what to block. CISOs need functional groupings that map to risk profiles and policy, not marketing copy.
Our taxonomy classifies every domain by what it does, not what it claims to be. Block Code & Development and you block every AI code generator — from well-known services to single-developer side projects that launched yesterday.
Each of the 18 top-level categories represents a distinct functional domain, with subcategories for selective enforcement. Block all of Text & Language while allowing Translation & localization, or block Code & Development except for approved Code review tools.
Every domain gets one of 18 categories and one subcategory. This maps directly onto firewall URL-category rules and DNS policy groups — no transformation needed.
Domains are classified by what the AI feature does, not by the company behind it. A productivity suite's AI writing assistant goes under Text & Language — tools cannot hide behind a benign parent category.
The taxonomy evolves as the AI landscape changes — we added Agents & Automation when autonomous agents emerged as a distinct class. Subcategories are added, merged, or split as tool populations shift.
Click any category to expand its subcategories. Counts reflect the current number of classified AI-tool domains in each bucket.
A binary block-or-allow approach rarely matches organizational reality. Marketing may need Marketing, Sales & SEO tools while legal requires everything else blocked — and engineering needs Code & Development while finance must be fully restricted.
Subcategories add a second dimension of control. Allow Translation & localization while blocking General assistants & chatbots, or permit Music generation while hard-blocking Voice cloning for deepfake prevention.
On top of the 18 functional categories, every domain in the database carries four security-risk flags — individually verified against each tool’s own site, terms of service and corporate ownership:
Consumer/free tiers that use submitted content for model training under their own ToS — the classic shadow-AI leak channel.
Ownership, hosting or jurisdiction that places submitted data under foreign data-access laws — including wrapper sites that pipe data to those model operators.
AI porn generators, adult chatbots and explicit companion platforms — a separate flag from abuse, essential for schools and workplace policy.
Deepfake and “nudify” tools targeting real people, uncensored/jailbroken chatbots, and AI tooling for phishing and fraud.
The combination is what makes policy possible: “allow productivity copilots, block anything that trains on our data or sits under foreign jurisdiction” is one rule against two columns. The flags also feed our five-dimension AI risk-assessment framework; the full risk methodology lives on the high-risk blocklist page.
Every record includes category and subcategory — whether delivered as CSV, API JSON, or EDL. Here is what a sample record looks like across all supported formats.
# domain, category, subcategory, status, language, ai_type chatgpt.com,Text & Language,General assistants & chatbots,active,en,genai midjourney.com,Image & Visual,Text-to-image,active,en,genai github.com/features/copilot,Code & Development,Code assistants & autocomplete,active,en,genai
{
"domain": "claude.ai",
"is_ai_tool": true,
"category": "Text & Language",
"subcategory": "General assistants & chatbots",
"status": "active",
"last_checked": "2026-07-09"
}
# AI Tools Blocklist — Text & Language category # Generated: 2026-07-09 | Count: 8,423 domains chatgpt.com chat.openai.com claude.ai gemini.google.com perplexity.ai writesonic.com jasper.ai copy.ai ... (8,415 more domains)
Download the free sample, explore the full database, or contact us about custom category feeds and OEM licensing.
Tell us which categories or subcategories you need and we will deliver a targeted domain feed in your preferred format.