On February 13, Sergey Saburov, Singapore Foundation Ambassadors Team Lead at Acronis, delivered a guest lecture at Nanyang Technological University (NTU) in Singapore to over 50 undergraduate students.

The session, titled “AI Security Risks: Real-World Examples,” explored practical attack vectors such as prompt injection, instruction hijacking, and vulnerabilities emerging from orchestration logic rather than the model itself.

Rather than focusing solely on model internals, the lecture reframed AI security as a system-design challenge – emphasizing trust boundaries, interface assumptions, and how real-world deployments introduce new risk surfaces. The goal was to bridge theoretical concepts with real production security considerations.

One student reflected:

The lecture significantly reshaped how I think about LLM safety… it was particularly thought-provoking to consider that many vulnerabilities emerge from orchestration logic rather than the model itself.

This kind of engagement highlights the importance of bringing real-world security experience into the classroom and helping future engineers build more secure AI systems from the ground up.