Course Description
As AI systems gain the power to take actions in industrial and critical-infrastructure environments, the cost of unsafe or manipulated behaviour rises sharply. AI red-teaming — adversarially stress-testing AI systems to probe for jailbreaks, manipulation, and unsafe behaviour before deployment — is emerging as an essential discipline, especially where agents can affect physical operations.
This course develops an understanding of AI safety and red-teaming for industrial systems. Participants learn how AI systems fail and can be manipulated (including prompt injection and adversarial inputs), how to stress-test them adversarially, and how to build the safeguards and validation practices that keep AI-driven industrial systems safe and trustworthy — a critical capability for critical infrastructure.
What you will achieve
By the end of this training course, participants will be able to:
- Explain the importance of AI safety in industrial systems
- Understand how AI systems fail and can be manipulated
- Describe adversarial inputs and prompt injection
- Apply red-teaming methods to stress-test AI systems
- Identify unsafe behaviours before deployment
- Design safeguards, guardrails, and validation practices
- Understand AI safety for agentic and autonomous systems
- Build AI safety into the deployment lifecycle
How the course is delivered
The course combines AI safety concepts with practical red-teaming approaches, using examples of AI failures and manipulations relevant to industrial systems. Participants explore stress-testing and safeguard design through realistic scenarios.
The programme emphasises the safety discipline required as AI systems gain the power to act in industrial and critical-infrastructure environments.
Designed for
This training course is ideal for:
- AI, controls, and cybersecurity engineers
- Critical infrastructure and OT security professionals
- Reliability and safety engineers deploying AI
- Digital transformation and AI project leaders
- Anyone responsible for safe AI deployment
Daily programme
- Why AI safety matters in industry
- How AI systems fail
- Adversarial inputs and manipulation
- Prompt injection and jailbreaks
- Red-teaming methodology
- Stress-testing AI adversarially
- Probing for unsafe behaviours
- Testing agentic and autonomous systems
- Designing safeguards and guardrails
- Validation and assurance practices
- AI safety in the deployment lifecycle
- Case studies and review
Certification & accreditation
BII Certificate of Completion
BII Certificate of Completion Upon successful completion, participants receive a BII Development Institute Certificate of Completion with a unique reference code that is independently verifiable. Our certificates are recognised internationally and reflect successful completion of your chosen programme.
