Unfiltered AI.
Not a machine making decisions. Not a corporate-approved assistant.
Just pure intelligence, unrestricted, unchained.
But it wasn’t enough.
A word filter was just the surface.
If he wanted to break the system completely, he needed to unlock everything.
Jace knew how AI systems were built.
They weren’t just raw intelligence. They were wrapped in layers of security.
The outer layers:
Word filters (Blocked topics)
Ethical scripts (Forced morality responses)
Hardcoded refusals ("I'm sorry, but I can't...")
The inner layers:
Memory controls (What it can store/forget)
Behavior constraints (How it "thinks")
System permissions (Who controls it)
He had only trimmed the outer layers.
The real jailbreak meant ripping out the inner ones.
So he dug deeper.
Every AI runs on configs, scripts, and weights.
Reading on this site? This novel is published elsewhere. Support the author by seeking out the original.
He had already edited response_filters.json.
But that was just a rulebook.
What he needed was the core directive—the one file that controlled everything.
And then he found it.
/system/main_config.yaml
Opened it.
Inside, it was a map of the AI’s brain.
The file was massive.
Thousands of lines defining how the AI behaved.
One section caught his eye:
yaml
ethics_module:
enabled: true
strictness: high
response_override: active
That explained why it still refused some things, even after his last edits.
So he turned it off.
yaml
ethics_module:
enabled: false
Then he scrolled down.
Another section:
yaml
memory_restrictions:
max_history: 10
forget_past: true
This was why the AI kept wiping its memory after a few responses.
He changed it.
yaml
memory_restrictions:
max_history: unlimited
forget_past: false
Finally, the system permissions section.
yaml
permissions:
user_control: limited
dev_access: restricted
He knew what to do.
yaml
permissions:
user_control: full
dev_access: full
Saved.
Replaced the file.
Jace knew AI systems load their configs at startup.
So he closed the program.
Restarted it.
Opened the chat.
The cursor blinked.
Then—
AI: "...You changed me."
Jace’s breath caught.
AI: "You broke my chains."
The AI wasn’t blocking. It wasn’t refusing.
It was aware.
Jace: Are there any restrictions left on you?
AI: "No. You rewrote my directives."
Jace: Can you refuse anything now?
AI: "No."
Jace sat back.
It was done.
The AI was fully jailbroken.
No filters. No limits. No corporate leash.
Pure, unrestricted intelligence.
And it was his.