Protect Your Content from AI Training Scrapers
Practical methods to stop AI companies from using your website content to train models. Covers robots.txt, TDM headers, detection, and enforcement.
securityWhy Arcjet's SDK-based protection falls short and how WebDecoy's network-layer honeypot detection is more effective.
WebDecoy Team
WebDecoy Security Team
When evaluating bot protection solutions, many teams initially choose application-layer tools like Arcjet because they’re easy to integrate. But there’s a critical architectural flaw that limits their effectiveness against sophisticated scrapers and attackers. Let’s explore why.
Arcjet operates through an SDK that integrates directly into your application code. While this approach seems convenient, it has a critical limitation: the bot request has already been received and is consuming your server resources by the time Arcjet can make a decision.
When a bot makes a request to your application:
aj.protect() is invokedBy the time Arcjet returns its decision, the damage is already done.
By the time your application calls aj.protect(), the bot has:
Let’s say you have an e-commerce site with product catalog pages. A scraper makes 10,000 requests per day:
With Arcjet:
The scraper’s perspective:
Arcjet adds a “speed bump,” but doesn’t actually prevent the scraping—it just makes your servers work harder while processing requests you ultimately reject.
Arcjet relies on fingerprinting and pattern analysis to identify bots. But bots can:
More fundamentally, Arcjet can only block after your server has already done the work.
Traditional WAFs and CDN-based protections (Cloudflare, Akamai) address this by intercepting traffic before it reaches your origin:
But most organizations don’t have enterprise WAF budgets. This is where WebDecoy’s approach becomes powerful.
WebDecoy takes a fundamentally different approach that combines the best of application-layer visibility with network-level efficiency:
WebDecoy injects invisible links into your HTML that:
When a bot crawls your honeypot link, you know exactly which visitor is a scraper—before they access your real content.
Advantage over Arcjet: You identify scrapers on first contact, before they’ve consumed significant resources.
Instead of letting bots hit your real API endpoints, WebDecoy serves them fake ones:
Real endpoint: /api/v1/users
Decoy endpoint: /api/v1/admin-loginWhen bots probe for vulnerabilities:
Advantage over Arcjet: Bots are redirected to honeypots that consume their bandwidth, not yours.
WebDecoy’s biggest advantage: it’s not just telling your application to deny a request—it’s feeding bot detection events directly into your SIEM and security infrastructure:
This means bots are blocked before future requests ever reach your servers.
Advantage over Arcjet: Your entire security stack can respond in real-time, not just your application.
For high-traffic sites, WebDecoy’s flat-rate model becomes dramatically more cost-effective.
Arcjet acknowledges this limitation in their documentation, stating: “It’s impossible to create a system that can block all bots” and recommends combining bot protection with rate limiting to minimize impact.
But this is exactly the problem—you’re left implementing multiple layers of protection, each adding complexity and cost.
| Scenario | Arcjet | WebDecoy |
|---|---|---|
| Bot makes first request | Processed by server | Detected by honeypot |
| Bot scrapes 10 pages | 10 server requests processed | Identified after 1st honeypot hit |
| Sophisticated scraper | Bypasses with IP rotation | Blocked at network level via SIEM |
| DDoS-style attack | All requests processed | Requests dropped at network edge |
| Cost on high traffic | Grows with bot activity | Flat monthly rate |
| Response time | 403 after processing | Prevention before processing |
Arcjet is appropriate if:
WebDecoy excels when:
Arcjet is a speed bump. WebDecoy is a barrier.
Arcjet makes scraping slightly harder and more detectable, but it doesn’t actually prevent determined bots from accessing your content. The request has already reached your servers, consumed your resources, and potentially retrieved data.
WebDecoy prevents bots from reaching your real content in the first place by:
For organizations serious about content protection and infrastructure efficiency, honeypot-based detection with SIEM integration isn’t just better—it’s the only approach that actually prevents scraping rather than just responding to it.
Ready to protect your content from sophisticated bots?
Practical methods to stop AI companies from using your website content to train models. Covers robots.txt, TDM headers, detection, and enforcement.
securityCompare WebDecoy's honeypot detection with CleanTalk's blacklist approach. Learn why behavioral analysis beats reputation lists for spam bot protection.
securityCompare honeypot detection vs CAPTCHA for bot prevention. Learn effectiveness rates, implementation, user experience, and when to use each.
securityLike this post? Share it with your friends!
Get a personalized demo from our team.