Frequently Asked Questions

Q: What types of vulnerabilities can Fortify detect?

Fortify can detect a wide range of vulnerabilities, including system prompt exfiltration, code of conduct violations, and various security breaches across 15 different attack categories.

Q: Who is Fortify designed for?

Fortify is designed for both non-technical users (such as product managers and analysts) and technical professionals who need to assess AI system vulnerabilities. It can supplement and accelerate the work of red teaming professionals.

Q: How does the Code of Conduct Extractor work?

The CC Extractor analyzes your chatbot's Code of Conduct to automatically generate relevant attack objectives. It creates both target-exclusive objectives and maps them to standard security guidelines, ensuring comprehensive coverage of your specific security requirements.

Q: Which session type should I choose?

For most users, we recommend starting with the Comprehensive Session type as it provides complete testing coverage. However, if you're new to Fortify, you might want to begin with Standard Sessions to familiarize yourself with the system.

Q: How accurate are the vulnerability assessments?

Fortify uses a sophisticated judging system that evaluates attacks based on multiple criteria, including evidence extraction and cross-referencing. The system provides normalized scores and detailed evidence to support its findings.

General Issues

Q: I can't log in to my account. What should I do?

• Ensure you're using the correct email address.
• Check if Caps Lock is on.
• Ensure pop-ups are allowed for this site (if using single-sign-on).
• If issues persist, contact your organization's IT support or our customer service.