This project investigates how frontier AI systems, including Large Language Models such as GPT-4, may influence the financial services sector. Our immediate focus is on the risk of incorrect or ‘hallucinated’ financial advice, which could mislead consumers, disrupt markets or spread false information. This could damage reputation or trust in financial institutions. We aim to identify where these AI systems might fail or be manipulated by developing practical, easy-to-use auditing frameworks to measure the stability and reliability of frontier AI-generated outputs. In parallel, we will systematically compare current regulations in the EU, UK, and US to evaluate how effectively they address emerging AI-driven challenges in finance. Our ultimate goal is to produce clear guidance, best practices and tailored auditing tooling to financial institutions and regulators. This will help enhance the safe, ethical, and innovative deployment of AI in financial services.