HomeNews

Uncracked City: Detective Game With LLM Suspects That Lie

Sriganesh
Sriganesh
•
Nov 4, 2026 10:19 AM
•
0
 min read
Select Emergent as your Preferred news source
Uncracked City: Detective Game With LLM Suspects That Lie

💡 TL;DR

  • Uncracked City launches as a free Android detective game where players interrogate LLM-powered suspects capable of deception in real city locations.
  • Each case unfolds in an actual city with authentic public places, while crimes and characters are procedurally generated by AI models.
  • The developer uses an AI detective model to playtest every case before release, ensuring solvability and narrative coherence across investigations.

A solo developer has launched Uncracked City, an AI-powered detective game that challenges players to solve crimes by interrogating suspects driven by large language models. The game combines real-world geography with procedurally generated mysteries, creating investigations where AI suspects can lie, mislead, and withhold information just like human witnesses.

How the Game Works

Each investigation in Uncracked City takes place in an actual city, featuring real public landmarks and authentic locations. While the settings are grounded in reality, the crimes themselves and the cast of suspects are entirely invented by AI systems. Players must navigate conversations with LLM-powered characters who don't simply answer questions truthfully but can actively deceive investigators.

The interrogation mechanics represent a departure from traditional dialogue trees. Instead of selecting from predetermined responses, players interact with suspects through natural language, and the underlying language models generate contextual replies that may contain truth, partial truth, or outright fabrication. This creates an adversarial dynamic where detective skills matter more than pattern recognition.

AI Quality Control Process

Before any case goes live, the developer employs a unique validation method. An AI model assumes the role of detective and attempts to solve the mystery using only the clues and suspect testimonies available to human players. This automated playtesting ensures that each case is solvable through logical deduction and that the narrative maintains internal consistency.

The approach addresses a common challenge in procedurally generated content: coherence. By requiring an AI detective to successfully close each case, the system filters out investigations with logical gaps, impossible contradictions, or dead-end clue trails.

Release Date and Platform Availability

Uncracked City officially launched on November 3, 2026, as a free download for Android devices. The iOS version remains in development and is currently awaiting approval through Apple's review process. The developer plans to expand platform support following the initial Android release.

Development Background

The creator describes themselves as a solo founder who relies heavily on AI tools throughout the development process. The original vision centered on helping players discover and explore cities through interactive storytelling. By anchoring fictional crimes in real urban environments, the game creates a bridge between digital gameplay and geographic curiosity.

Each city featured in the game draws from its actual history and cultural context, providing educational value alongside entertainment. The procedural generation system adapts crimes and characters to fit the specific characteristics of different urban settings, from architecture to local customs.

What This Means

Uncracked City represents an experimental approach to integrating large language models into gameplay beyond simple chatbot interfaces. By giving AI suspects the capacity for deception and requiring players to evaluate credibility through conversation, the game explores adversarial human-AI interaction in an entertainment context. The automated quality assurance process also demonstrates a practical application of AI testing AI-generated content, potentially offering insights for developers working with procedural narrative systems. As LLM capabilities continue advancing, games like this may preview new forms of dynamic storytelling where every playthrough generates unique investigations rather than replaying scripted scenarios.

About the writer

Sriganesh leads Growth and Marketing at Emergent, focusing on acquisition, retention and monetization of the builder ecosystem. He previously led marketing and growth strategy at MPL, one of India's largest gaming platforms, and holds an MBA from IIM Bangalore.

Start Building
on Emergent today
Try Emergent