AI Errors Create Reputational Risks, Says Workiva

A new study from Workiva found that one in four organisations have detected AI errors that reached external audiences or board members. The finding comes despite 84% of business leaders reporting confidence in AI outputs without human oversight, according to Workiva's 2026 Midyear Executive Benchmark Survey.

The gap between executive confidence and operational reality could create challenges for marketing teams responsible for protecting brand reputation and managing stakeholder relationships. Automated content tools are being deployed faster than the governance systems designed to oversee them, leaving corporate disclosures and market communications exposed to unverified information.

The survey gathered responses from 2,272 finance, risk and sustainability professionals, including 847 C-level executives across North America, Latin America, Europe and the Asia Pacific region. Workiva also evaluated feedback from 367 institutional investors based in North America and the United Kingdom.

Data quality blocks deployment

Speaking to Finance Chief, Junko Swain, Chief Accounting Officer at Workiva, says: "Workiva's Midyear Executive Benchmark report surfaces a clear pattern: as leaders deepen their partnership with AI, their confidence grows, but so does their awareness of its risks. In fact, 26% of executives said internal AI audits have detected errors that reached external audiences or the board.

"This figure demonstrates that AI errors aren't theoretical risks. Human oversight and data lineage are baseline requirements. This is especially true in fields like internal and external financial reporting, accounting processes and sustainability disclosure, where errors carry regulatory, reputational and legal consequences."

The study indicates that weak data architecture is preventing organisations from scaling automated tools across business functions. According to the data, 27% of leaders state that poor data quality has blocked AI deployment within core workflows.

A further 71% report that data deficiencies have had at least a moderate negative impact on AI usage across financial and sustainability reporting. Only 11% of executives believe their organisation's current data quality is sufficient for AI application.

Investor confidence at risk

The challenge extends beyond internal operations to external market relations. The survey revealed that 89% of institutional investors are concerned about the accuracy of AI-generated content in corporate disclosures.

For marketing leaders, this could signal new pressure on brand credibility and communications accuracy. Organisations that fail to address data deficiencies could risk damaging stakeholder relationships and losing competitive ground.

Junko says: "Poor data generates wrong answers, and often wrong answers that look entirely credible. I believe leaders are expressing confidence in the controls and review processes around it. That's a reasonable starting point, but only if those controls actually exist and are operating effectively. In too many organisations, AI adoption has outpaced the governance framework, and no single person is accountable for closing that gap.

"Without clear ownership, it's too easy for the hard questions to go unasked: who validated this output? Where did the data come from? Is this auditable?"

Control systems become priority

Barbara Larson, Chief Financial Officer at Workiva, says: "Confidence in AI without control over data quality is a liability, not a strategy. CFOs need platforms that connect AI to trusted, auditable data so every output is one they can verify and every disclosure is one they can defend.

"Getting this right is about more than avoiding errors. Business leaders can move faster and embed AI deeper into their operations when they trust what their systems produce. That's a real competitive edge."

To address these risks, organisations are calling for infrastructure to oversee automated agents. Rather than relying solely on broad AI models, leaders are prioritising tools that ensure traceability, governance and continuous audit capability.

Survey respondents outlined several infrastructure requirements:

• 55% cited platforms to manage autonomous agents and automated workflows
• 49% pointed to traditional systems of record such as general ledgers
• 45% highlighted software that enables full auditability

Junko says: "When those errors do get caught, it's often because an experienced professional noticed something that didn't look right. That's valuable, but it's not a system. A governance model that depends on someone happening to spot the problem is a governance model that will eventually miss one.

"For finance leaders, confidence in AI must extend all the way down to the data layer. That means knowing precisely where data originated and whether it can withstand scrutiny."