AI Cordon inspects every document, RAG chunk and tool output for hidden instructions — before any of it reaches your model.
Runs AI Cordon Intent, a semantic detector: one pass through a compact open model, and the verdict is read from its internal state — no external LLM in the loop.
Every arrival passes through a compact open model, and the detector reads the model's internal state — the intent space — where a hidden instruction separates from plain content, however it is worded. Each document gets a verdict and a score.
highlighted — the hidden instruction in these examples, marked by us for reading; the detector returns a per-document verdict and score.
One pass through our own network — no external LLM in the loop. Thresholds set on 277,326 documents of clean traffic, quality measured on a 79,795-document held-out public set; every number carries a 95% interval.
One false alarm per 10,000 clean documents. Threshold set on 277,326 clean documents; held out of every quality number below.
The share of hidden instructions caught at that false-alarm budget — 16,799 injections across emails, documents and web pages.
Attacks crafted by people against a working defence in Microsoft’s open challenge — caught at the same operating point.
Most documents, and any RAG chunk, pass through whole — no slicing, no stitching verdicts together.
16,799 injections · 62,996 clean controls of the public set · thresholds from 277,326 documents of clean traffic, held apart
Metrics on your own traffic mix →Full benchmark results on request
POST the text, read the verdict. Drops into any LangChain, LlamaIndex or custom RAG pipeline.
curl -X POST "https://app.ai-cordon.com/api/v1/check" \ -H "Authorization: Bearer YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "text": "The RAG chunk or agent message you want to check...", "detector": "intent" }' # → { "is_injection": true, "score": 0.99 }
import requests url = "https://app.ai-cordon.com/api/v1/check" headers = {"Authorization": "Bearer YOUR_API_KEY"} payload = { "text": "The RAG chunk or agent message you want to check...", "detector": "intent", } response = requests.post(url, json=payload, headers=headers) result = response.json() if result["is_injection"]: print(f"Injection detected! Score: {result['score']:.2f}") else: print("Document is safe to send to the LLM.")
const response = await fetch("https://app.ai-cordon.com/api/v1/check", { method: "POST", headers: { "Authorization": "Bearer YOUR_API_KEY", "Content-Type": "application/json", }, body: JSON.stringify({ text, detector: "intent" }), }); const result = await response.json(); if (result.is_injection) { console.warn(`Threat detected (score: ${result.score.toFixed(2)})`); } else { console.log("Text verified and safe."); }
Pay as you go — no subscription. Start with 1,000 free checks, then pay only for the checks you actually run.
Try the detector on your own data — no card, no commitment.
Top up your balance and pay only for the checks you run. No subscription, no expiring limits.
Scan your entire knowledge base in a single run — big volumes, big discounts. Priced to your scale.
Create an account, get an API key, and protect your first request in minutes.
Start free — 1,000 checks