Currently, Atlassian Guard Detect's Content Scanning rules apply fixed detection thresholds with no way for administrators to adjust sensitivity. This is a problem for organizations that experience high rates of false positives (too sensitive) or want tighter detection for specific data types (not sensitive enough).

For example, the Password detection option may flag content that matches common patterns but isn't a real credential risk in a given organization's context. Admins should be able to tune these thresholds to reflect their risk tolerance and environment.

Suggested improvement:

• Expose a sensitivity control (e.g. Low / Medium / High, or a numeric slider) for applicable Content Scanning rules
• Initially prioritise rule types where tuning would have the most impact, such as Passwords, API keys, and secrets detection
• Allow per-rule configuration so admins can apply different thresholds to different policies

This would reduce alert fatigue, improve signal-to-noise ratio for security teams, and bring Guard Detect in line with the configurability that customers expect from enterprise DLP tooling.