The AI Confidence Gap: 84% Trust the Output, 26% Have Seen Errors Reach the Board

But the same survey reveals that 26% — one in four executives—reported that internal audits have detected AI errors that reached external audiences or board members . This gap between stated confidence and documented failure is the central finding of the report.

Workiva CFO Barbara Larson put it directly: "Confidence in AI without control over data quality is a liability, not a strategy" .

Data Quality Is the Weakest Link

The confidence gap sits atop a more fundamental problem: data quality. Only 11% of executives believe their organization's data quality is sufficient for AI use . The remaining 89% acknowledge that their underlying data is not ready for the AI tools they are already deploying.

This mismatch has real consequences. 71% of executives said poor data quality has at least moderately impacted the use of AI in financial and sustainability reporting . The top reported consequences of poor data quality are bad or delayed operational decisions, followed by regulatory fines or legal action and the loss of investor or lender credibility .

Investors Are Watching Closely

Institutional investors are not reassured by executive confidence. 89% of institutional investors said they are worried about the accuracy of AI-generated content in corporate disclosures . This concern comes alongside growing regulatory pressure to treat AI outputs as auditable records rather than informal drafts .

95% of institutional investors believe leaders underestimate the risk of fragmented financial reporting data . Over time, weak or unverifiable data may erode credibility with stakeholders .

Infrastructure Priorities: The Path Forward

Executives are responding with a clear infrastructure mandate. The survey found that 96% of executives said C-suite alignment is imperative to break down data silos , and 79% are prioritizing data automation and governance to close enterprise-wide data gaps . Most organizations are funding this shift with dedicated IT support (73%) and allocated budgets (71%) .

76% of internal audit teams are already actively evaluating their organizations' AI models . Business leaders almost unanimously agreed that the CFO, CIO, and CSO must unite for data governance to succeed .

The infrastructure priority is moving from "execution to orchestration" . CFOs are specifically seeking platforms that serve as a single "system of truth"—grounding AI agents in traceable, regulatory-grade data environments where every output can be verified and every disclosure defended .

AI Agents Enter the Reporting Workflow

In parallel with the survey findings, Workiva launched three purpose-built AI agents designed to operate within auditable guardrails: a Tie-Out Agent for financial report consistency checks, a Benchmarking Agent for peer disclosure analysis, and a Sustainability Disclosure Agent for compliance with ESRS and ISSB standards . These agents are embedded in a persistent intelligence layer called Workiva Knowledge, which grounds AI output in an organization's own data, instructions, and content .

The audit-trail architecture matters because AI-generated outputs in regulated financial filings are now a named risk category in FINRA's 2026 Annual Regulatory Oversight Report . Hallucination risk is highest where AI has the most latitude to generate narrative—management commentary, footnote disclosures, and ESG language—which is exactly the territory these agents are designed to keep grounded in source data .

The Bottom Line

Workiva's 2026 Midyear Executive Benchmark Survey paints a picture of an industry moving fast on AI adoption but moving too slowly on the data foundations and governance structures needed to make that adoption safe. The numbers are stark: 84% trust, 26% errors caught, 11% data-ready, 89% investor concern. Closing the gap between AI confidence and AI control is now the defining challenge for finance, risk, and sustainability leaders.