Can You Break This AI? AI Red Teaming, Testing & Risk in Practice

Event Confirmation Form!
➡️ We kindly request that all attendees complete the event confirmation form to confirm your attendance:
https://meet.zoho.com.au/hbcq-ufc-btv

***

## What’s On?

Speaker: Sandeep Bhalekar, Founder – Global Institute of Artificial Intelligence

Title: Can You Break This AI? – AI Red Teaming, Testing & Risk in Practice

Abstract: As organisations deploy increasingly capable AI systems and AI agents, asking “Does the AI work?” is no longer enough.
The real questions are: How can it fail? Can it be manipulated? What happens under unexpected or adversarial conditions? And can we discover those weaknesses before they create business risk?
This practical session introduces AI red teaming, testing and evaluation, exploring how organisations can deliberately challenge AI systems to uncover vulnerabilities, unexpected behaviours and failure modes before and after deployment. NIST defines AI red teaming as structured testing designed to identify flaws and vulnerabilities in AI systems, including undesirable behaviours and misuse risks.

## What You’ll Learn

✅ AI Red Teaming Explained — Understand what red teaming is and why organisations deliberately challenge their AI systems.
✅ Finding AI Failure Modes — Explore hallucinations, unsafe outputs, bias, unexpected behaviours and control failures.
✅ Prompt Injection & Agent Hijacking — Understand how malicious instructions can manipulate AI applications and agents. NIST's 2026 agent-security work specifically highlights indirect prompt injection and agent hijacking as risks for agents processing external data.
✅ Testing AI Agents — Explore the additional risks when AI can access data, call tools and take actions.
✅ Risk-Based AI Testing — Learn how testing can be prioritised around system purpose, business impact and potential harm.
✅ From Testing to Governance — Turn discovered weaknesses into guardrails, controls, monitoring and continuous improvement.
✅ Pre-Deployment + Continuous Testing — Understand why testing before launch alone is insufficient and why deployed AI also needs monitoring.

## Who This Is For

AI leaders, CIOs, CTOs, AI engineers, developers, cybersecurity professionals, Responsible AI practitioners, risk and compliance teams, AI governance professionals, consultants and anyone involved in building, testing, deploying or governing AI systems.
No specialist red-team experience is required.

## Format

90-minute practical live session featuring AI failure scenarios, red-team thinking, testing approaches and real-world examples, followed by an open Q&A.
The session is built around one simple challenge:

Before your customers, employees or attackers
find the weakness in your AI — can you find it first?

Please complete the confirmation form above to confirm your attendance.