Open Models Almanac
Methodsevaluationcybersecurity - vulnerability discovery; exploit development

single mention · evaluation

Cybersecurity guardrails

Model safeguards that distinguish defensive cybersecurity tasks from malicious or offensive ones.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See cybersecurity - vulnerability discovery; exploit development for the methods filed beside it.

Also called built-in guardrails.

Filed under evaluation › cybersecurity - vulnerability discovery; exploit development