EP.26 — The Crime That Never Happened
Tokyo | May 12, 2045
The police arrived before anyone was hurt. No weapon had been found. No victim had called for help. No crime scene existed. Yet officers quietly escorted a middle-aged office worker from a crowded train station after the National AI Prevention Network issued its highest-level warning. According to the system, the probability of a fatal attack within the next forty-three minutes had reached 99.97 percent.
Authorities insisted they had not arrested the man for what he might do. Instead, they described the intervention as a temporary public safety measure based on millions of real-time signals—financial stress, online activity, biometric data, travel patterns, emotional changes, and environmental conditions.
The man denied any criminal intention. His lawyer argued that dangerous predictions could never replace human actions. Thoughts were not crimes, and probabilities were not evidence.
Three days later, investigators discovered a detailed attack plan hidden inside an encrypted cloud account. The suspect admitted he had considered carrying it out but claimed he had not yet made a final decision. Prosecutors responded that waiting for certainty would have meant waiting for someone to die.
The Supreme Court delivered a landmark ruling. It declared that preventive intervention could be lawful when an AI system demonstrated overwhelming predictive reliability, provided that the action remained temporary, transparent, and subject to immediate judicial review.
Civil rights organizations warned that society had crossed a historic line. For centuries, justice punished completed crimes or attempted crimes. Now, it had begun responding to futures that existed only as statistical certainty.
The verdict left one question echoing around the world: If technology can stop almost every crime before it happens, will freedom be measured by the risks we are still allowed to take—or by the dangers that algorithms no longer permit?
— Joonghwa Byeon
