Indirect Prompt Injection via Web Content Processing Leads to Agent Hijacking and Data Exfiltration | AI Breaking Wire