For people who use AI, the right way

AI makes mistakes.
Crith catches them.

Crith checks every AI answer in real time, right in your browser. Because AI is always right, right?

ChatGPT
Should I put my savings in index funds?
Where should we begin?
CRITH Missing caveats

Assumes a 10+ year horizon, an emergency fund already in place, and tolerance for 30%+ drawdowns. “Safest” is highly conditional.

One extension. Every AI chat.
Add to Chrome How it works →
1 in 3
AI answers contain errors
Real-world benchmarks find roughly a third of AI responses include hallucinated facts, unsupported claims, or missing context.
66%
Of people trust AI without questioning it
Most users accept AI answers at face value, even when the stakes are real.
100%
Of answers audited
Crith reads every AI response automatically. You only see a flag when something's actually worth your attention.
Works on
How It Works
01
CRITH reads every AI response
When you ask ChatGPT, Claude, or Gemini a question, CRITH analyzes the answer in the background — automatically.
Best way to learn Spanish in 3 months?
02
It flags what's worth checking
Missing context, unjustified assumptions, confident claims with no source, things that might be hallucinated — all surfaced in plain English with the exact phrase dotted-underlined.
Yes, low-cost index funds are the safest long-term investment for most people. Steady 8-10% returns over time.
CRITH Missing caveats
Assumes a 10+ year horizon and tolerance for 30%+ drawdowns. “Safest” is highly conditional.
03
You decide what to do
Push back on the AI, verify a specific claim, or move on. CRITH gives you the second opinion you didn't know you needed.
Push back on the AI
Verify a specific claim
Move on
Your call. Every time.
Why CRITH

Don't trust
AI blindly.

AI sounds confident even when it's hallucinating. CRITH audits every response and flags what's worth checking — so you trust the answers that matter and verify the ones that don't.

01 · AI is a yes-man by default
It validates what you already think and says what sounds right, not what is right. Crith catches that and shows you what to question.
02 · Hallucinations are real
AI confidently invents citations, statistics, and quotes. Crith flags claims that are worth verifying before you rely on them.
03 · Works where you already work
ChatGPT, Claude, Gemini, Perplexity, Grok, DeepSeek. Crith audits AI everywhere you use it. One extension, all your tools.
Get Started

Stop trusting
AI blindly.

Takes 30 seconds to install. Crith audits every AI response — flags hallucinations, unbacked claims, and missing context, so you don't trust the wrong answer.

Add to Chrome
Chrome Web Store · Free to install · Fully private
FAQ

How CRITH audits AI

Common questions about auditing AI responses and catching what's wrong.

CRITH is a Chrome extension that audits AI responses. When you use ChatGPT, Claude, Gemini, or other AI chats, CRITH analyzes each response in the background and flags what's worth questioning — missing context, unjustified assumptions, claims that might be hallucinated. The exact phrase that triggered the flag is dotted-underlined in the response, with a small CRITH logo next to it.
ChatGPT, Claude, Gemini, Perplexity, Grok, and DeepSeek. One extension, all of them.
No. CRITH runs in the background after the AI finishes responding. The AI experience is unchanged — CRITH adds a thin audit layer alongside it. If the response is solid, you see nothing.
Yes. AI responses are sent to our backend for analysis only. We never sell your data and never use it to train AI models. You can delete your account and all associated data anytime. Full details on the privacy page.
CRITH doesn't replace your AI tools. It audits the ones you already use. ChatGPT and Claude don't tell you when they might be wrong. CRITH does.
Your prompts are classified privately. Never sold or used to train AI.
Only installs from the official Chrome Web Store.
Delete your account and all data anytime.