Imagine building a house with a smart lock that can be tricked by a simple sticker. This project creates a 'security check' for AI to make sure it can't be fooled or biased. It also builds smarter tools that help programmers find bugs automatically without giving too many wrong alarms.
AI components introduce new vulnerabilities like poisoning and bias that traditional security tools cannot detect. Simultaneously, AI-driven security tools produce too many false positives for companies to use in official certifications.
A set of open-source tools and methodologies for AI security testing, including benchmarking data and runtime monitoring techniques.
If you are an aerospace manufacturer dealing with autonomous safety systems—this project developed assurance methods that ensure AI components are robust and safe for flight. This reduces the risk of system failure in critical aviation environments.
If you are a 5G provider dealing with core network virtualization—this project developed runtime monitoring techniques that detect threats and correct misconfigurations. This prevents exploits in critical digital infrastructure.
If you are a certification body dealing with 3rd party software assessment—this project developed benchmarking data and certification processes. This allows for a standardized way to verify that AI-augmented software is actually secure.
Based on available project data, no specific pricing or cost information is provided as the project focuses on generating open-source tools and methodologies.
Yes, the project validates its approach through real-world pilots in 5G, aviation, and software certification, involving large enterprises like Airbus, SAP, and Thales.
The project aims to generate a set of open-source tools, though specific licensing agreements for the 13 partners are not detailed in the provided text.
It develops certification methods and assurance processes for AI/ML components to align with the European Digital Resilience and Sovereignty strategy.
The tools are designed to support DevOps teams in secure coding and vulnerability mitigation, specifically targeting the reduction of false positives in AI-powered security tools.
The consortium is heavily weighted toward industrial application, with a 46% industry ratio comprising 6 companies, including global giants like Airbus, SAP, and Thales. This balance of 5 universities and 4 SMEs ensures that the research is grounded in commercial reality and targeted at high-value sectors like aerospace and telecommunications.
Contact Universita Degli Studi di Trento
Contact us to explore the open-source tools for AI security certification.
