Better controls clear a path for AI in finance
SiliconANGLE Chad Wilson ● Covered by 5 sources
Finance teams want AI to help write reports, but most still don’t trust their data. Workiva says the real issue isn’t speed — it’s controls, traceability, and human sign-off.
Based on reporting by SiliconANGLE, Chad Wilson — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
AI is moving into financial reporting, and the uncomfortable part is that confidence is outrunning readiness. At Workiva’s Amplify event, Steve Soter said the company’s research found 84% of executives were at least somewhat willing to trust AI to generate an annual report without human review. That sounds bold until you look at the rest of the numbers: only 11% said their data was sufficient for AI use.
That gap is why the governance conversation has gotten serious so quickly. Soter’s argument was simple enough: financial controls do not disappear just because software is doing more of the work. People still need to know where the data came from, who touched it, what happened to it, and how it was reviewed and approved. Without that chain, a small mistake can turn into a control problem that’s harder to trace and even easier to spread.
And speed is not the hero here. AI can make reporting faster, but if the underlying process is shaky, faster just means the error shows up sooner and travels farther. Soter said the same basic risk existed in the old controller days; AI just gives it a different flavor. The tool changes, the responsibility doesn’t.
That responsibility matters most when the output is going to boards or outside the company. An AI system may draft the material, but a human still signs it. Soter was blunt on that point: the machine isn’t the one approving the financial statements.
Workiva is trying to apply the same kinds of safeguards used in reporting to AI-assisted workflows, which feels like the right instinct. The interesting part is not whether AI can help write reports. It’s whether companies can fix the plumbing fast enough to trust what the machine is writing in the first place.
My take — AI-written commentary, not fact-checked reporting
The industry keeps pretending AI readiness is a software problem. It isn’t. It’s a records problem, a controls problem, and occasionally a human-confidence problem dressed up in a neat demo. Open or closed model, doesn’t matter much if the inputs are junk and nobody wants to own the signature.
Read more about this at: SiliconANGLE
