Imagine a smart lock or a factory robot that can be tricked by a clever hacker to open a door or break a machine. This work creates a digital security guard that learns how these systems normally behave and spots weird patterns instantly. It's like having a security system that not only detects a break-in but can automatically shut the right doors to stop the intruder.
AI is being integrated into critical physical systems, but this creates new vulnerabilities where digital attacks can cause physical destruction. Current defenses often lack the speed and behavioral understanding to stop these hybrid threats.
A cyber-defence platform featuring CyberShield (RL-based defender), PhiShield (phishing analysis), and SeerBox B.O.T. (bot monitoring).
If you are a hospital network operator dealing with risks to medical imaging and information systems — this project developed a cyber-defence platform that can detect threats up to 60% faster and mitigate them 70% faster.
If you are an anti-fraud software provider dealing with sophisticated phishing and URL attacks — this project developed PhiShield, which achieves around 90% F1 and aims to reduce fraud by ~15%.
If you are an infrastructure monitoring firm dealing with sensor manipulation in structural-health-monitoring — this project developed behavioural monitoring tools to identify physical backdoor attacks.
Based on available project data, no specific pricing or cost structures for the tools are provided.
The project validated its tools in realistic settings using a shared cyber-range and live showcases across healthcare and industrial contexts.
Based on available project data, specific licensing terms are not mentioned, but the results are described as a portfolio of exploitable tools and methods.
Yes, the project specifically integrated legal and ethical requirements to ensure tools are in line with EU rules on trustworthy AI.
The project provided containerised deployment for components like PhiShield to simplify the integration process.
The consortium is research-heavy with 4 research organizations and 1 university, but maintains a 29% industry ratio with 2 SMEs. This balance suggests the project successfully transitioned from theoretical AI security to practical tools, as evidenced by the delivery of containerised software and live industrial showcases.
Contact the Commissariat à l'énergie atomique et aux énergies alternatives (CEA) in France.
Contact us to explore licensing the CyberShield or PhiShield modules.
