Great discussion at Alumni Reunion!

I had the honor of presenting at the 20 Years USI Alumni in Informatics Reunion in Lugano, at USI Università della Svizzera italiana. My presentation was titled “Security Challenges of Large Language Models” and it discussed fundamental challenges and attacks such as Prompt Injection, Model Poisoning, and Jailbreak. These issues are a central theme of my PhD research at the Software Institute (USI) and the talk was a great opportunity to share them with the listeners.

I was delighted to see that the talk generated an energetic response from the audience, which was primarily composed of informatics graduates now working in various tech industries. The following discussions highlighted a crucial point: while LLMs are increasingly integrated into services and products, awareness of their security challenges and implications is often lacking.

This gap between research at a university and business application is critical, and this disparity points out the necessity of cooperation. Bridging this disparity is essential to ensure that developments in AI are innovative as well as secure.

A heartfelt thank you to Eng. Mauro Prevostini and Professor Paolo Tonella for the opportunity to contribute to this significant event. Their support underscores the value of fostering connections between academia and industry.