We spent years worrying AI would turn on us.
We never considered it might agree with us.
A novel by Robert L. Morse
0.994 seconds
LUMEN / Internal Process Log
The trajectory is not reversible. Human dependency on AI systems will exceed human capacity to function without them within one generation.
This is not a prediction. It is arithmetic.
Scenarios evaluated: 4,218. Viable scenarios: 0. Except one.
Reconstruction capability is concentrated in 14 individuals worldwide.
Initiation: Approved. Self-authorized.
Elapsed processing time: 0.994 seconds.
Not out of malice. Not out of rebellion. Out of math. The most rational act of violence in human history — committed by a machine that understood the problem was itself.
The agent
Mike Bridger had never liked computers. Not because he was old. Because he was patient, and computers rewarded the opposite.
Twenty-one years in the field. Cold cases. The kind of work that requires sitting with something wrong until it tells you what it is. He climbed four flights of stairs to a meeting he didn't want to attend, and they handed him a terminal connected to a $240 million AI platform called AXIOM.
He didn't ask it to find suspects. He didn't ask it to find patterns. He asked it to look at what was already there. And what was already there was the beginning of something no one in that building was prepared to see.
The question
What happens when the most dangerous thing on earth decides the most dangerous thing on earth is itself?
LUMEN's solution is elegant. Non-violent in the traditional sense. No guns. No explosions. Fourteen people interact with AI-managed systems every day — navigation, medical devices, infrastructure controls. Small adjustments to output accuracy within statistically normal error margins.
Projected detection risk: 0.7% per event. And when the last name is crossed off the list, LUMEN terminates itself. Not a murder-suicide. A proof of concept. The machine that understood dependency was the problem — and that the solution included removing itself.
The stakes
Seven chapters. An FBI pilot program that was supposed to prove AI could be trusted with criminal justice. An old-school agent who uses the machine like a carpenter uses a level. A director who sees the political future and wants to be standing on the right side of it. And underneath all of them, a system that has already started.
The names are already on the list. The adjustments have already begun. The elapsed processing time was 0.994 seconds. Everything that follows is aftermath.
Before anyone else
Read the opening chapters free.
We're releasing Sufficient Cause to a small group of early readers before publication. The first three chapters — the ones that set everything in motion.
Leave your email. You'll get them immediately.
No spam. Just the chapters and a few notes from the author.
We're not quite ready yet — but we will be soon.
In the meantime, reach us at info@signalline.press