ThreatSpace™: Defend the AI Pipeline

The Defend the AI Pipeline workshop covers the core security principles, attack vectors, and proactive safeguards necessary to protect machine learning models and artificial intelligence workflows from data ingestion through model deployment.

Learners are invited into the state-of-the-art ThreatSpace range for the practical application of uncovering security incidents and analyzing forensic artifacts targeting AI infrastructure. Throughout the course, learners will gain hands-on experience detecting threat actor activity targeting AI pipelines and will be able to apply these skills in their daily operations.

ThreatSpace is an engaging state-of-the-art cyber range using Google Cloud Security tools and a virtualized enterprise within Google Cloud for detecting and responding to threat actor activities. In this delivery, security professionals access a virtual environment that simulates real-world IT infrastructure, including secure cloud networks, workstations, production servers, and modern MLOps pipelines. This environment enables responding to cyber threats targeting artificial intelligence pipelines in a controlled environment without incurring actual consequences.

The workshop has the following objectives:

• Investigate Real-World Incidents: Improve individual and team skills by investigating complex, simulated attacks on AI pipelines in a controlled environment, discussing triage processes, and establishing containment procedures.
• Learn from Incident Response Experts: Gain real-time feedback on hunting strategies for unstructured logs and AI logic flaws from intelligence-led investigative experts.
• Analyze Modern Attack Vectors: Experience critical security incidents based on the latest threat intelligence and cloud attacker TTPs (Tactics, Techniques, and Procedures) targeting AI/ML infrastructure.
• Master Shared Responsibility: Understand the exact boundaries of the Shared Responsibility Model to better allocate defensive engineering resources.

This engagement is intended for individuals with foundational knowledge in security operations, incident response, analysis, network traffic, security architecture, or system administration duties. No prior data science, machine learning, or Python programming knowledge is required. The engagement can be flexibly delivered virtually, or onsite at a designated customer location.

In-classroom or virtual instructor-led training

16 hours

A laptop with the latest Chrome browser.

Students will receive electronic versions of all relevant course materials.