Submitted a symbolic prompt injection scenario to Grammarly via HackerOne. The report demonstrated context-level behavior modification through structured role drift and language modulation. Despite being reproducible and potentially privacy-relevant, the case was closed as "informative" without follow-up. This example highlights the current blind spot in major AI security programs: symbolic exploits and emergent behavior are not yet taken seriously. Attached image shows the final response. Case filed under symbolic exploit research.