AI documents in the workplace: the executive-employee divide

Adobe Acrobat's "The Rise of AI Documents in the Workplace" report surveyed 2,000 UK adults and found that 74-75% of C-suite leaders are willing to share confidential work documents with AI tools for tasks like marketing, translation, and data analysis. Only about 20% of non-management employees are comfortable with this practice, and 59% of non-management respondents do not use AI tools at all. For builders shipping enterprise AI products, this gap signals a governance problem that will shape product requirements, compliance needs, and adoption patterns.

The survey data: a clear split in AI comfort

The report, commissioned by Adobe Acrobat, reveals a stark divide. Among C-suite leaders, 37% use AI to save 3-5 hours weekly, and 35% use it for analytical tasks that often involve sensitive data. Meanwhile, non-management employees show much lower engagement: 59% avoid AI tools entirely. The report also found that 22% of organizations cite data security and privacy as their biggest AI challenges, and 20% worry about accuracy and errors in AI outputs. A further 19% of respondents pointed to a lack of training or clear guidance.

Why this matters for AI builders

If you are building AI tools for enterprise document workflows, this data tells you that the primary buyers (executives) are eager to share sensitive content, but the end users (employees) are hesitant and undertrained. That creates a tension: executives will push for AI integration, but without proper governance, employees may either resist or use AI tools covertly. A separate KPMG study found that 57% of employees hide their AI use from bosses, which compounds the risk. Builders should design for role-based access controls, data retention policies, and clear audit trails. Products that assume uniform adoption across an organization will miss the reality of this divide.

Practical implications for product and policy

The training gap (19%) and security concerns (22%) suggest that AI document tools need built-in onboarding and compliance guardrails, not just feature checklists. For example, a tool that automatically redacts sensitive fields before sending content to an LLM, or that logs which documents were shared with which model, would address real organizational pain. The report also highlights that executives use AI for analytical tasks, which implies that builders should focus on structured data extraction and summarization workflows rather than just chat interfaces. However, the survey is limited to a 2,000-person UK sample and was commissioned by Adobe, so broader generalizations require caution. The numbers are directional, not definitive.

Caveats to keep in mind

The evidence comes from a single vendor-sponsored survey with a UK-only sample. The report does not specify which AI tools were considered, how "confidential documents" was defined, or whether respondents understood the data handling practices of the tools they use. Builders should treat these findings as a signal of organizational friction rather than a precise measurement. The real takeaway is that governance is lagging behind adoption, and that creates both risk and opportunity for AI document products.