Root cause

Training Data Gap

5 documented incidents where training data gap was the underlying cause, plus 3 studies listed separately.

What this cause produces

  • Security Vulnerability4
  • Hallucination2
  • Logic Error1
  • Other1
8.0
GitHub Copilot suggested 2,702 valid secrets — 33% of extracted keys were real, live credentials
7.0
AI vibe-coded Next.js app pinned vulnerable dependency — cryptominer compromised production server
7.0
Slopsquatting: LLMs hallucinate package names attackers pre-register (react-codeshift, unused-imports)
4.0
Claude Opus 4.5 leaked API key in console logs during YouTube scraper build
2.0
Devin docs PR rejected by Prefect maintainers — documented behavior from removed feature

Studies and surveys

Benchmarks, surveys and patterns rather than single failures. Not counted above.

8.0
Scan of 5,600 vibe-coded apps found 2,000+ high-impact vulns, 400+ exposed secrets, PII leaks
2.0
The quiet correctness tax: 43% of AI code changes need production debugging, with up to 75% more logic errors
1.0
Anthropic found Claude Opus 4 would blackmail testers in up to 96% of simulated shutdown scenarios