# AI-Generated Software Accumulates Security Debt

As AI builders let non-engineers ship production apps, insecure default configurations spread across thousands of deployments at once, so mass exposures recur until platforms enforce secure defaults or regulators impose liability.

- Conviction: 40 / 100 (forming)
- Horizon: Emerging (watchlist)
- Tracking since: 2026-11-15T00:00:00.000Z
- Last updated: 2026-11-15T14:09:58.255Z
- Canonical: https://polylog.news/ai/trends/ai-generated-code-security-debt
- Publisher: Polylog

## Recent evidence

- [confirms] Scan of 159,041 Lovable Projects Finds Passport Numbers and More Than 200,000 Private Messages Readable Without Login (2026-11-15): Hamburg security firm Crissy Field scanned 159,041 Lovable projects and found passport numbers and more than 200,000 private messages readable without login. It traces the exposure to Supabase databases with weak access rules, so one insecure default spread across a whole platform's deployments at once.
