Pyrodit Security · security research
Make the basis for security decisions visible
Programme overview
A security check should explain what it examined and which evidence supports its finding. We study how text scanning, sensitive-data masking and permission decisions can use these findings with clear responsibilities.
- Find risk signals using rules, name recognition and analysis of meaning
- Scan reports findings; Guard applies organisational rules to decide what happens next
- Masking hides sensitive text while keeping a record of the transformation
Research plan · no measured results published yet
What we plan to evaluate
- How many real risks are detected and how often harmless content is incorrectly flagged
- How much of the requested check completed, was skipped or failed
- Typical response time, slow-request response time and memory use
- Impact on harmless tasks and consistency under small text changes
- Whether results can be measured again with the same check version and test data
Define the scope
Record which text, language and check version will be used. Set time and resource limits before the check runs.
Record findings and their basis
Keep the risk signal, its location in the text and the portion actually examined together. Content that could not be checked must not be reported as safe.
Apply the organisation’s rules
Scan reports the finding. Masking hides or transforms the affected text. Guard applies the organisation’s rules to allow, block or request human approval for an action.
Make later review possible
Record the check version, the applied rule and the resulting action together so the basis for a decision remains auditable.
About this report
These are technical research reports, not peer-reviewed publications. Each report identifies its source, measurement conditions and limitations. The evidence section explains what can be downloaded or reproduced.
Request the research materials →