Enterprise AI Agents Face Security Hurdles Despite Leadership Optimism

A SnapLogic survey released this week reveals a paradox in enterprise attitudes toward AI agents: while IT decision-makers largely back the technology, significant concerns about its deployment are emerging. The poll of 1,000 IT leaders found that roughly 85% believe AI agents can match or exceed human worker performance on assigned tasks. Four out of five respondents indicated that IT departments stand to gain the most from rolling out these systems.

Yet barriers to widespread adoption loom large. Data security and privacy rank as the top concern among respondents, trailed by difficulties integrating legacy systems, insufficient staff knowledge, and anxiety over AI hallucinations producing unreliable outputs.

Investment and Real-World Deployment

Agentic AI has become the technology sector's current focus, with vendors actively promoting solutions and organizations showing genuine interest. According to SnapLogic's findings, nearly 80% of IT leaders intend to allocate at least $1 million toward AI agent initiatives in the current year. Most plan to deploy a dozen or more agents within the following 12 months.

Beyond IT operations, other business units are experimenting with the technology. Walmart has deployed an AI agent to assist merchants in diagnosing supply chain problems, as executives noted during a February earnings presentation. Marketing teams at Accenture are leveraging agents for campaign development. Toyota Motor Corporation has implemented agents to support internal knowledge management functions.

Security and Governance Risks

Industry analysts caution that organizations pursuing agent deployments without establishing robust governance and security frameworks risk serious consequences. Gartner has forecasted that one in four enterprise security breaches over the next three years will stem from AI agent misuse. Forrester projects that three out of four companies attempting to build agents internally will struggle and ultimately seek assistance from outside consultancies.

Infrastructure and Workforce Readiness

Most enterprises remain in early preparation stages. A Tray.ai survey determined that nearly 9 in 10 IT professionals report their organizations require technology infrastructure improvements before they can successfully implement AI agents.

Organizational change management will prove essential to successful AI adoption. A Pegasystems report found that approximately 2 in 5 employees express discomfort with submitting work generated by AI systems, citing concerns about output quality falling short of their own standards. Workers also harbor doubts regarding the technology's dependability, precision, and the transparency of how it operates.