Hackers, Killers and Defence Systems

Our immune system protects us from invading viruses, bacteria and parasites. But it also defends us against an enemy from within: cancer. The problem is that cancers develop ways to hide, and to suppress our natural defenses. In our lab, we design precision molecules that allow our immune system to more effectively find and attack cancer cells.

T cells are immune cells which help keep viral infection and cancer under control. It takes only an incredibly tiny abnormality in the proteins being displayed to the immune system for a T cell to recognise danger and initiate a potent immune response, however the exact mechanism which gives the T cell this power is not yet fully understood. Our goal is to once and for all explain how the T cell kills with such precision and sensitivity, to then apply these killer instincts for therapeutic good!

In 1945, Oppenheimer built the most dangerous weapon in history with slide rules and chalkboards. What could he have done with ChatGPT? We'll find out, and discover the safety and security pitfalls of modern AI along the way. We'll break its safety guardrails, catch it lying with confidence, and trace how these same problems show up in the AI systems you already use every day.

Artificial intelligence is being rapidly adopted across personal, industrial, and even military domains. Often, this adoption comes with overlooked cybersecurity risks, many of which are unique to AI systems. This talk explores how AI systems might be hacked, demonstrating real attack strategies and the kinds of damage they can cause in the real world.