AIAI Seminar-Monday 25 March 2024-Talk by Tarini Saka and Bhargavi Ganesh

Speaker: Tarini Saka

Title: A Collaborative Human-AI Approach to Mitigate Large-Scale Phishing Attacks

Abstract: Phishing is a form of cyber-attack in which fraudulent emails, text messages, or other electronic communications are sent to trick recipients into revealing sensitive information, such as login credentials, credit card numbers, or social security numbers. Given the low cost of operation and high potential rewards, phishing is becoming increasingly popular. In recent years, phishing often occurs in the form of campaigns, a coordinated effort to target a large number of individuals or organizations with phishing emails. This has led to an exponential increase in the number of phishing attacks and emails, causing a massive burden on organizations. Therefore, it is crucial to implement an efficient system to promptly address phishing attack reports and mitigate potential risks. My research aims to develop an innovative intelligent system that enhances the effectiveness of existing phishing mitigation systems used by organizations. Leveraging the latest advancements in Artificial Intelligence (AI) and Natural Language Processing~(NLP), I aim to design a system that employs advanced grouping techniques to swiftly identify phishing campaigns on the back end to support IT Help teams. Furthermore, in terms of user interaction, our proposed system will enhance the capabilities of auto-responders by equipping them with comprehensive information to create well-informed reports when responding to users who report emails.

Speaker: Bhargavi Ganesh

Title: Evaluating the Global AI Governance Landscape

Abstract: As AI has continued to be used in a wide range of areas, such as health, education, social media, etc, there have been growing concerns regarding the ability of existing governance mechanisms to hold actors accountable for the impacts of AI systems. Recent developments related to the increased accessibility of generative AI models have further accelerated calls for AI-specific policy proposals worldwide. In addition to formal governance methods, such as regulations and legislative proposals, there have also been a number of informal strategies pursued, such as the development of organizational management-related standards and guidelines like the ISO 42001 and NIST Risk Management Framework. In my research, I have developed a taxonomy to classify the different approaches being pursued, particularly focusing on efforts being undertaken in the US, UK, and EU. In this talk, I will first present this taxonomy, and a brief analysis of the political economy factors shaping the development of global AI governance. Then, I will discuss some of the practical implementation challenges that are likely to arise within each of the categories in this taxonomy. Finally, I will conclude with some ideas regarding directions for technical research that can help support more empirically-grounded AI policymaking going forward.