OpenAI used a recent webinar to demonstrate how its finance team uses AI to speed up forecasting and close processes. But the audience's questions focused on something the presenters largely avoided: how to keep that AI under control.
The presenters emphasized faster workflows, automation, and collaboration. Throughout the event, the chat filled with questions about data governance, internal controls, SOX compliance, and auditability. Few of those questions received direct answers during the Q&A session.
Presenters focused on speed, attendees asked about safety
Kyle Kober, director of product finance at OpenAI, Stephanie Struck, head of product finance, and Jackson Wang, a data scientist on the technical staff, showed how ChatGPT Work automates forecasting, reconciles financial models, and builds interactive dashboards. Wang described AI making forecasting "a lot more fun" by replacing solitary spreadsheet work with live scenario collaboration. "Make forecasting great again, guys," he said.
Meanwhile, participants asked who controls data access, where the data is stored, and how the workflows align with existing information technology general controls. The disconnect between the demo and the chat illustrates a central obstacle to enterprise AI adoption in finance: vendors are selling speed, but buyers are vetting trust.
The governance gap remains wide
Kober did describe ChatGPT Work as "governed and auditable with controls to lock down the spend and the usage." The presenters also mentioned administrator-controlled access to Sites and the use of approved "golden tables" as trusted data sources. For most finance leaders, those references were too general to satisfy deeper concerns about control design and effectiveness.
The demand for a reliable data foundation is backed by hard numbers. An Intuit survey found that 70% of finance leaders lack a single source of truth for critical business data. More than half said delayed visibility caused them to miss a strategic opportunity in the previous six months. Gartner has argued that organizations should aim for a "sufficient version of the truth" rather than a perfect one. As finance teams evaluate AI tools, many are starting with data architecture and governance before moving to models - a common focus in an AI for Finance learning path.
Confidence in governance lags behind deployment speed
The tension in the webinar mirrors broader market data. A Deloitte survey found that 93% of large organizations now use AI across multiple functions, but only 40% of CFOs said they are "very confident" in their organization's AI governance framework. More than half of CFOs cited a lack of governance authority as an obstacle, and the most common concern was balancing pressure to deploy AI quickly with the need to manage risk.
Those same worries surfaced directly in the OpenAI chat. Finance leaders submitted question after question about who is responsible for the data the AI uses, the costs it incurs, and the systems governing it. The AI Learning Path for CFOs increasingly treats governance maturity as a prerequisite for deployment, not an afterthought.
Why this matters for finance professionals
The OpenAI webinar confirmed that the conversation around AI in finance has shifted. The technology works. The question now is whether it can be trusted. Finance professionals who lead the push for clear guardrails, audit trails, and control frameworks will determine whether AI scales without creating material risk for their organizations. Getting the governance right is not a blocker to innovation - it is the foundation that makes responsible innovation possible.
Your membership also unlocks:
