Two years after financial institutions rushed to adopt generative AI, the industry is now questioning what must be in place around the technology before it can be trusted, according to Tearsheet editor Sara Khairi. In 2023, the focus was access, as banks experimented with ChatGPT-like tools after the technology demonstrated human-like language understanding, but this created unrealistic expectations that AI could eventually understand finance. By 2024, the industry shifted from replacing workers to assisting them, with Morgan Stanley’s OpenAI partnership integrating GPT-4 into tools to help financial advisors search research more efficiently, without managing client relationships or making investment decisions. Banks and fintechs launched their own ChatGPT-like assistants and enterprise LLM deployments for employees and customers to search information, draft content, and streamline work, with the most successful implementations reducing friction. By late 2025 and into 2026, the conversation moved to governance, data, and ROI, as institutions began building AI agents designed to execute specific financial tasks like underwriting, compliance, and servicing on top of proprietary systems.