Back to the library

Key question New

AI-processed · Author-approved

Why do formal training and policies fail to ensure safe generative AI use, and how can companies fix it?

Activating Employee’s AI Literacy for Responsible Generative AI Use: A Sensitization Framework
Michael Dominic Harr, Leonardo Banh, Reinhard Schütte, et al. · International Conference on Information Systems (ICIS) · 2026

11 views

Illustration for: Why do formal training and policies fail to ensure safe generative AI use, and how can companies fix it?

Loads Spotify's player; Spotify may set cookies. This device remembers your choice.

Key findings

  • Identified four core dimensions of organizational AI sensitization: organizational factors, individual factors, sensitization practices, and structural tensions.
  • Revealed four structural barriers explaining why responsible AI use fails even among knowledgeable staff: adoption-engagement gaps due to workload, knowledge-practice disconnections leading to shadow AI, heterogeneity misalignment from one-size-fits-all formats, and identity-efficiency tensions.
  • Showed that teaching employees about generative AI capabilities without providing capable internal tools can paradoxically increase the use of unauthorized external tools under time pressure.
  • Introduced the Sensitization Profile Matrix classifying employees into four archetypes: Uninitiated, Naïve Adopters, Cautious Gatekeepers, and Responsible Innovators.
  • Established that promoting responsible AI adoption requires structural enablement—such as dedicated work time, competitive governed tools, and peer champions—rather than purely educational lectures.

What it means for you

CIO / IT Executive

Audit current enterprise GenAI tooling against external commercial alternatives to ensure internal tools are fast and capable enough to stop employees from resorting to shadow AI under workload pressure.

IT Manager

Identify and empower tech-savvy peers within your team to act as localized AI champions, moving away from one-size-fits-all training to address specific daily workflow bottlenecks.

Business Strategist

Formally allocate dedicated 'exploration and safe practice' time into project schedules so employees are not forced to bypass AI guidelines purely to meet tight deadlines.

Researcher

Design empirical studies that investigate how organizational time pressures and identity conflicts directly trigger shadow AI usage, moving beyond purely cognitive compliance models.

Policymaker

Draft regulatory frameworks that incentivize organizations to provide structural enablement and competitive internal tools, rather than relying solely on mandatory compliance lectures and punitive guidelines.

Summary

This study examines how organizations can activate employees' knowledge to foster responsible generative AI adoption rather than relying on one-off training. The authors conducted 20 semi-structured interviews with human resource professionals across 13 European organizations and analyzed the data using the qualitative Gioia methodology. Based on these insights, they developed a diagnostic matrix categorizing employee behavior and outlining targeted intervention pathways.

Problem

While organizations increasingly mandate guidelines and formal training for generative AI, employees frequently bypass official systems and use ungoverned tools to meet heavy workloads. Standard compliance programs fail to account for how day-to-day organizational pressures, technological shortcomings, and identity conflicts undermine safe AI practices.

Transcript
Host: Welcome to A.I.S. Insights — powered by Living Knowledge. I'm Anna Ivy Summers. Today, we are discussing a study presented at the International Conference on Information Systems 2026, titled "Activating Employees’ AI Literacy for Responsible Generative AI Use: A Sensitization Framework," authored by Michael Dominic Harr, Leonardo Banh, Reinhard Schütte, and Gero Strobel. The study explores how organizations can move beyond ad-hoc training to genuinely embed responsible generative AI use into daily organizational practice. Joining me is our analyst, Alex Ian Sutherland. Welcome, Alex.

Expert: Thanks, Anna. Glad to be here.

Host: Alex, many companies today mandate compliance guidelines or offer one-off training modules, yet employees still struggle with responsible adoption or turn to ungoverned tools. What core problem does this study address?

Expert: The authors observed a persistent governance gap. Organizations often introduce formal policies, mandatory e-learning, or proprietary tools. However, when faced with high workloads and tight deadlines, employees frequently bypass these official frameworks. They use personal, ungoverned AI accounts—commonly referred to as shadow AI—to complete their tasks. The study highlights that relying solely on individual training or rigid rules fails to resolve the practical workplace pressures that drive this behavior.

Host: To examine how organizations can bridge this gap, what approach did the researchers take?

Expert: They conducted twenty semi-structured interviews with human resource professionals across thirteen European organizations. HR represents an informative setting because these professionals handle sensitive personal data under strict regulations like the GDPR and the European Union AI Act, while also managing workforce upskilling. The researchers analyzed the interview data using the qualitative Gioia methodology to inductively identify key dimensions, practices, and structural tensions.

Host: What were the primary findings from this analysis?

Expert: The study established four aggregate dimensions: organizational factors, individual factors, sensitization practices, and structural tensions. Crucially, the authors identified four structural barriers that explain why responsible AI practices often fail, even among staff who understand generative AI.

Host: Could you walk us through those four barriers?

Expert: First is the adoption-engagement gap. Sensitization programs exist, but work intensification and operational pressures prevent employees from finding time to participate. Second is the knowledge-practice disconnection. When sanctioned internal tools lag behind public alternatives in speed or features, employees use their AI knowledge to bypass corporate safeguards. Third is heterogeneity misalignment, where standardized, one-size-fits-all training fails to address varying levels of technical skill and role-specific demands. Finally, the identity-efficiency tension emerges when using generative AI conflicts with professional identity, such as values around authentic human communication or contextual judgment in recruitment.

Host: That finding regarding shadow AI is notable—simply teaching employees about AI capabilities can actually encourage workarounds if internal tools fall short. How does the study suggest managers evaluate these dynamics?

Expert: The authors developed the Sensitization Profile Matrix, which maps employees across two dimensions: AI literacy and actual generative AI usage. This outlines four profiles: the Uninitiated, who have low literacy and low use; Naïve Adopters, who use AI frequently without assessing risks like hallucinations or data privacy; Cautious Gatekeepers, who understand the risks and deliberately limit their use; and Responsible Innovators, who combine regular adoption with critical evaluation.

Host: Looking at the business implications, what actionable steps should leaders take based on these insights?

Expert: The central takeaway is that responsible AI adoption requires structural enablement rather than just didactic training. Organizations need to provide dedicated, manager-sanctioned time for employees to experiment. Additionally, if companies want to prevent shadow AI, their governed internal tools must be functionally competitive with external options. Finally, organizations should replace uniform e-learning modules with tailored, role-specific learning paths supported by peer champions who demonstrate responsible practices directly within daily workflows.

Host: That provides a clear perspective on moving from basic compliance toward sustained organizational sensitization. Alex, thank you for sharing these insights with us.

Expert: My pleasure, Anna.

Host: And thank you to our listeners for tuning into A.I.S. Insights — powered by Living Knowledge. Join us next time for more analysis at the intersection of business and technology.
Artificial Intelligence & Automation Information Systems & Management Human-Computer Interaction Data Analytics & Big Data Digital Transformation & Strategy Cybersecurity & Privacy Cloud & Infrastructure E-Commerce & Digital Platforms Operations & Supply Chain Sustainability & Green IS

Generative AI, AI Literacy, Responsible AI, Shadow AI, Human Resource Management, Organizational Governance