One platform. Every AI modality. Local execution.
Adversarial testing and robustness evidence for your Computer Vision, LLM, RAG, and Agentic AI models. Local execution. Technical reports for your audit files.
Built on research, ready for production
research-based
by design
PyPI
European Union
EU AI Act
One suite. Four domains. One CLI.
Whatever your model, RednBlue covers the full AI spectrum with a single command.
Vision Robustness
Adversarial tests for classifiers and object detectors. Gradient masking diagnostic.
LLM Security
Jailbreak tests, prompt injection, token manipulation, RAG attacks.
Agent Security
Tool hijacking, goal drift, privilege escalation for autonomous systems.
Compliance Reports
Technical evidence supporting your EU AI Act (Art. 15), NIST AI RMF and ISO/IEC 42001 files.
All modules accessible via a single unified CLI: pip install rednblue. Pick what you need, pay per use.
In 3 commands
Works with your stack
Whatever your AI frameworks and providers, RednBlue integrates seamlessly.
ML Frameworks
LLM Providers
Agents & RAG Frameworks
Your model weights stay with you
Unlike cloud platforms, RednBlue never accesses your model's weights, architecture, or training data: it all stays on your infrastructure. For Vision testing, we receive encrypted metrics, each image's top-5 predictions, image hashes and a hashed model alias, never the images or the weights. For LLM, RAG, and Agent testing, the encrypted result also includes the test probes and brief response excerpts needed to substantiate each finding in your report.
- 100% local execution
- AES-256 encryption
- No access to model weights or architecture
- GDPR compliant
Technical evidence for your audits
Our reports provide technical evidence that supports the main regulatory frameworks.
EU AI Act
Article 15
NIST AI RMF
Measure 2.7
ISO/IEC 42001
AI Management
MAIGF
Singapore
RednBlue provides technical robustness evidence, not official regulatory certifications. Our reports serve as supporting documentation for your compliance files.
See It In Action
Watch how RednBlue tests a model in minutes
Built for modern AI teams
ML Freelancers
Ship your models with a professional robustness report. From €100.
See offer →AI Teams
Regular testing for your models before deployment. Signed, verifiable reports.
Discover →Compliance Teams
Technical evidence for your EU AI Act, NIST, ISO 42001 files.
Learn more →Researchers
Preferential academic access for your ML security research.
Contact us →Ready to test your model?
Create a free account. Tokens are issued by our team: contact us to get started.