64% of Employees Build AI Agents, 91% Grant Access to Sensitive Company Data
Event summary
- 64% of full-time employees surveyed have built their own AI agents.
- 91% of these agents have access to company data, including customer records and financial information.
- 11% of workers report their AI agent has exposed or mishandled sensitive information.
- 71% built their AI agent in under a week, with 89% taking an experimental approach.
- 88% of employee-built AI agents are client-facing, raising risks of data exposure.
The big picture
The rapid, decentralized adoption of AI agents by employees highlights a growing governance gap in corporate AI strategies. As non-technical workers increasingly build and deploy these tools, companies face heightened risks of data breaches and operational disruptions. The trend underscores the need for adaptive policies that keep pace with employee-driven AI experimentation, particularly in client-facing roles where exposure stakes are highest.
What we're watching
- Governance Dynamics
- How companies will adapt policies to manage the rapid, decentralized adoption of AI agents.
- Risk Management
- Whether current safeguards can mitigate the 11% exposure rate of sensitive data by employee-built AI agents.
- Productivity Trade-offs
- The pace at which companies will balance AI-driven efficiency gains against heightened security risks.
Related topics