AI Crimes

AI Crimes is an interactive code inspection game where developers spot real bugs, security flaws, and sloppy naming left by AI coding agents.

AI Crimes is an interactive game designed to train developers to detect errors made by AI coding agents. Players examine code snippets containing real, anonymized mistakes that range from misleading variable names to critical security holes. After reviewing the evidence, users submit an incident report detailing the line number, what went wrong, and why the issue matters. The platform evaluates the report, reveals the official verdict, and displays what the AI actually did. Supported programming languages include TypeScript, Python, Go, Java, Ruby, and C++.

AI Crimes at a glance

Best for: Software engineer
Runs on: Web
Pricing: 19 free cases available without signing up.
Released: October 2026
Category: AI coding tool
Website: solveaicrimes.com

Key features

Examine real, anonymized AI-generated code snippets
Identify errors ranging from naming issues to security holes
Submit reports detailing the line number and error impact
Receive instant verdicts on submitted reports
Track cases solved and clean streaks
Compete on community leaderboards
Supports TypeScript, Python, Go, Java, Ruby, and C++
Start investigating without signing up

Use it for

Practicing for technical interviews that evaluate AI code review skills
Learning to spot common errors generated by AI coding assistants
Improving code review precision across multiple programming languages

Alternatives to AI Crimes

Mantis — Automates vulnerability discovery and patching using AI agents.
Offense — It gives coding agents structured security evidence through CLI, MCP, and HTTP.
Metabob — Provides proactive debugging and enforces safe implementation patterns for AI tools.
Would You Merge — Practice spotting subtle AI-generated bugs and security flaws in Python and JavaScript.
ARTEX — Runs autonomous multi-agent penetration testing with a dual-graph architecture.
CodeGraph — It eliminates file crawling, reducing AI agent tool calls by 88% and cutting costs.
Promptfoo — It simulates real users to uncover agent and RAG vulnerabilities.
Anti Samcheonpo — Catches redundant AI tasks and blocks wasted agent spending.
CodeScan Artex — It uses multiple parallel AI agents to read source code and verify vulnerabilities.
Codacy — Enforces unified coding standards and security policies across the software lifecycle.
Crit — It keeps human feedback localized and precise for AI agent code iterations.
PullRequest — Combines AI risk analysis with vetted human experts to catch critical vulnerabilities.

All AI Crimes alternatives