Detects bursts of Cisco AI Defense activities that may indicate iterative prompt-injection probing or attempts to access restricted data within a web application. The rule triggers when, within a ten-minute window, either five or more distinct Cisco AI Defense event IDs occur for the same application and connection, or three or more occurrences of the same event ID. It focuses on events that match prompt injection, code detection, PII, or PCI guardrails. Allowed PII/PCI prompt events are excluded via Cisco AI Defense Allowed Sensitive Prompt. The search uses a Cisco AI Defense ingestion macro and aggregates on application_key and connection_key, counting total events, distinct event IDs, and the maximum number of identical event IDs. It surfaces details such as repeated_event_ids, policy names, and guardrail types, along with event actions (allow/block) and related context like application and connection names. The rule is designed to surface coordinated or repetitive probing attempts targeting GenAI-related capabilities or restricted data within a web application. Implemented in Splunk using the Cisco AI Defense macro, eventstats, and SPL binning to a 10-minute window; thresholds and window can be tuned via the how_to_implement guidance. The finding is intended to alert operators to potential automated or exploratory activity that could precede data exposure or privilege escalation via AI-enabled workflows.