All stories
Uncracked City Unusual AI Community reports

The suspects can lie: Uncracked City turns AI chat into a detective game

Ask your own questions, confront a statement with a clue and build a report in a free-to-play Android mystery game set in real cities.

By AITrending2 min read
Sources (4)

Why now A November 3 Show HN introduction and fresh English discussion put the game's free-form AI interviews in front of developers; its Android listing was last updated October 30, rather than launching today.
Uncracked City promotional map illustration with the words Every city keeps a secret and an aircraft route.
Publisher-supplied English trailer poster, used as a promotional illustration. It is not a screenshot of a played case or evidence that the AI dialogue works as advertised. Source: Uncracked City / BNT Technologies

What if catching an AI character’s lie were the point of the conversation? Uncracked City makes that its detective-game hook: search locations, question suspects in your own words and pin a clue against a statement that does not fit. The developer says its AI suspects respond in character, and some deliberately lie.

The project reached Show HN on November 3, with public replies comparing it to older detective games and questioning whether talking to unreliable LLMs sounds like leisure. That is a small, visible discussion, not evidence of mass adoption. The Android store lists an October 30 update; the new discussion should not be mistaken for a November 4 app launch.

Ask freely, then make the evidence fit

The English listing describes typed or spoken questions, voiced suspect replies and an evidence board. Players eventually file a report identifying the culprit and motive, with proof for each line; harder cases also ask when and where. There are three attempts. The listing explicitly warns that lying does not make a suspect guilty.

Real cities provide the geography, while the founder says the crimes and people are fictional. Five training levels lead into harder cases. The founder also says a model must solve a case before it goes live—a project quality-control claim, not an independently verified guarantee that every mystery is fair or every answer consistent.

Android is available through Google Play, where the game is described as free to play, with no account required and an optional microphone. The privacy policy describes a daily allowance of free questions and free/premium plans; it does not establish unlimited free play. In the Show HN post, the founder says the iOS version is still awaiting Apple review. A game-master MCP integration is a future ambition, not a shipped feature established by that post.

This is a hosted experience. The policy says questions and replies are kept with progress, and identifies Anthropic for suspect dialogue with OpenAI as a fallback and for other tasks, including transcription. AITrending has reviewed the English sources, not independently played through the cases. The appeal is the interrogation mechanic; coherent deception and a solvable mystery remain things players must judge in practice.

Related reading