Enhanced security with restricted reference domains
Prevent unwanted or unsafe references in the output of the AI
Lisette Vredenburg
Product Owner
Security and reliability are central to Vragen.ai. That’s why we are introducing ‘Restrict Reference Domains’. With this setting, you can determine which external domains are allowed for links and images in the responses generated by Vragen.ai. This helps prevent unwanted or unsafe references from ending up in the AI's output.
What the feature involves
Under Settings, you will find the option Restrict Reference Domains (beta). Once you activate this:
Vragen.ai will only accept links and images from domains you have explicitly specified;
all other references are automatically removed or marked as safe.
You enter the allowed domains as a list, one domain per line.
Why this is relevant
This feature helps to enhance the quality and safety of responses. By limiting which external sources can be included in generated output, you reduce the risk of incorrect, undesirable, or potentially risky references.
Background: Improvement following Ethical Hacker research
The introduction of Restrict Reference Domains follows directly from insights gained through an independent Ethical Hacker research conducted under the Cyber Security Pen Test certification of the CCV. Read more about it in the article: "How Vragen.ai performs under pressure from an Ethical Hacker".
