Indirect Prompt Injection Leading to Data Exfiltration in LLM-Powered Applications | AI Breaking Wire