---
title: Blocking AI Platforms
description: How to evaluate and implement AI platform blocking decisions using SeeLLM's what-if analysis.
audience: [human, agent]
type: guide
version: v1
stability: stable
lastVerified: "2026-08-01"
prerequisites: [/docs/evidence-model/, /docs/api/authentication/]
related: [/docs/guides/weekly-trends/, /docs/mcp/tools/]
artifacts: []
---

## Before You Block

Blocking an AI platform is a significant decision. Before acting, use SeeLLM's **what-if analysis** to understand the impact.

### Using the What-If Tool

Via MCP, ask your AI assistant:

> "What would happen if I blocked GPTBot from my site?"

Or via the API:

```bash
curl "https://api.seellm.link/api/analytics/whatif-block?domain=example.com&platform=gptbot&range=30d" \
  -H "Authorization: Bearer sk_live_..."
```

The analysis shows:
- How many visits you'd lose
- Estimated value impact
- Whether the platform drives referral traffic (real users) or just crawls

## When to Consider Blocking

| Reason | Recommended Action |
|--------|-------------------|
| Platform only crawls, never refers users | Consider blocking — low value, high resource cost |
| Platform drives referral traffic | Keep it — those are real users discovering your content |
| Excessive crawl rate causing performance issues | Rate-limit rather than fully block |
| Legal / licensing concerns | Block with `robots.txt` and verify compliance |

## How to Block

### Via robots.txt

The standard way to block AI crawlers:

```txt
# Block OpenAI's training crawler
User-agent: GPTBot
Disallow: /

# Block Anthropic's crawler
User-agent: ClaudeBot
Disallow: /

# Allow ChatGPT to fetch pages for users (referrals)
User-agent: ChatGPT-User
Allow: /
```

Note the distinction: `GPTBot` crawls for training, while `ChatGPT-User` fetches pages to answer user questions. You may want to block one but allow the other.

### Via Cloudflare Rules

For more granular control, use Cloudflare WAF rules to block by user agent, IP range, or ASN.

## Monitoring After Blocking

After implementing a block, monitor your SeeLLM dashboard to verify:

1. The blocked platform's visits drop to zero
2. No unexpected impact on other traffic
3. Referral traffic from that platform's users still works (if you only blocked the crawler)