Indirect Prompt Injection in AI Helpdesk via Unsanitized Web Content Leads to Privilege Escalation | AI Breaking Wire