AI Risk Health Check
Financial services · 45 employeesSeries A FinTech
Critical prompt injection caught 48 hours before enterprise pilot
Their support chatbot was leaking account hints through crafted prompts. We mapped 9 AI touchpoints, scored each against OWASP LLM Top 10, and gave engineering a ranked fix list the CTO could ship the same week.
- Risk score
- 34 → 78
- Critical findings
- 12 flagged
- Delivered
- 6 days