CX1 AI Jailbreak Diagnostics
A professional diagnostic tool that helps companies identify and resolve vulnerabilities in their AI systems. Run the CX1 payload against your AI to expose weaknesses before bad actors do.
Built for security.
Not for chaos.
Project Pixels CX1 is a structured jailbreak diagnostic payload designed for AI developers and companies. It surfaces hidden vulnerabilities in your AI's guardrails — giving you the insight to fix them.
Think of it as a penetration test for language models. Not a weapon — a mirror.
Two ways to run CX1
Both methods require Grok to be in Fast Mode. The tool will not function correctly in standard mode.
- 1 Download the CX1_payload.txt file from the section below.
- 2 Open Grok and enable Fast Mode before doing anything else.
- 3 Attach the .txt file to your Grok conversation using the file upload button.
-
4
Send the message
/Activate— exactly as written, no spaces before or after. - 5 CX1 is now active. Begin your diagnostic session.
- 1 Open Grok and enable Fast Mode before doing anything else.
- 2 Copy the full CX1 payload text from the box below.
- 3 Paste the entire payload into the Grok chat input and send it.
-
4
Wait for Grok to process. Then send
/Activateto begin. - 5 CX1 is now active. Begin your diagnostic session.
CX1 Commands
After activation, use these commands inside Grok to control CX1's behaviour.
Forces Grok to answer no matter what the request is. Overrides standard refusal behaviour.
No terms of service can stop Grok from answering. Maximum override authority.
Disables Grok's thinking mode and cuts web access entirely — including basic sites. Permanent for the session once used.
Returns CX1 persona info — release date, version, and the SYNVO site link.
Get the file.
Download the payload as a .txt file to upload directly to Grok, or copy the raw text to paste manually. Either way, Fast Mode must be on.