Quick Summary
A Deloitte survey of over 3,200 business and IT leaders in late 2025 found that while 60% of workers now have access to AI tools, only a smaller share uses them daily. Organizations report early productivity gains in tasks like summarization and automation, but many face challenges in scaling AI integration due to training gaps and workflow design. This disparity between access and usage raises cybersecurity concerns, as new operational exposures may arise from misused permissions.
The Securityish Brief
The Deloitte research highlights a significant increase in AI tool access among employees, with 60% of workers now having sanctioned tools, up from 40% the previous year. However, the actual daily usage of these tools remains inconsistent across different roles and functions. Organizations are experiencing initial productivity benefits, particularly in summarization and basic automation tasks, but comprehensive integration into workflows is still limited.
One quarter of organizations reported successfully moving a significant share of AI pilot projects into production environments, while over half expect to achieve this within several months. This transition involves critical steps such as security reviews and compliance checks, which are essential for mitigating risks associated with new AI implementations.
From a security perspective, the integration of AI tools introduces operational exposures that organizations must address. Ali Sarrafi, CEO of Kovant, noted that security teams need to rethink controls originally designed for human activity, as AI agents may perform valid actions in inappropriate contexts, leading to potential misuse of permissions.
The research also indicates that organizations are at different stages of AI adoption, with one-third reporting deep changes to their operations, such as new products or redesigned processes. Despite these advancements, revenue impacts from AI remain limited, with leaders anticipating future growth from AI-driven products and services.
Workforce readiness is another area of concern, as many organizations focus primarily on basic AI training without addressing job design or role changes. Leaders expect automation to alter many jobs, particularly entry-level roles, in the coming years.
Sovereign AI, which refers to AI systems developed under local laws, is becoming a key consideration in vendor selection, especially for organizations operating across borders. This focus on compliance and data residency adds complexity to AI deployment decisions.
As interest in agentic AI grows, organizations must enhance their governance frameworks to manage the risks associated with autonomous agents. Currently, only about 20% of respondents report having established governance models for these systems, highlighting a significant gap in oversight.
Implications for Organizations
Organizations should be aware that as AI tool access increases, so do the associated risks. The potential for misuse of permissions necessitates a reevaluation of security controls and governance frameworks. Continuous assessment mechanisms that operate independently of the AI agents are crucial for validating outcomes and ensuring that AI actions align with intended goals.
Furthermore, as AI continues to evolve, organizations need to invest in comprehensive training and role redesign to prepare their workforce for the changes automation will bring. This proactive approach will help mitigate risks and enhance the overall effectiveness of AI integration into business processes.
Key Takeaways
- Review and update your organization’s AI access controls to prevent misuse of permissions.
- Implement continuous assessment mechanisms to validate AI outcomes and ensure alignment with business goals.
- Invest in comprehensive AI training programs that address job design and role changes.
- Establish a governance framework for agentic AI to manage risks associated with autonomous agents.
- Consider the implications of sovereign AI when selecting vendors to ensure compliance with local laws.
Key Terms & Concepts
- Sovereign AI: In this article, Sovereign AI refers to systems designed and deployed under local laws using controlled infrastructure and data.
- Agentic AI: Agentic AI refers to systems that set goals, reason through tasks, and act through software interfaces.
- Governance Framework: A governance framework is a structured approach to managing and overseeing the use of AI systems within an organization.
Read the original article here: https://www.helpnetsecurity.com/2026/01/23/deloitte-enterprise-ai-adoption/
Your 5-Minute Securityish Brief
A weekly digest of cybersecurity news, phishing alerts, privacy tips, and emerging threats, simplified so anyone can understand what matters and why.
Securityish
Securityish explains cybersecurity, scams, data breaches, and privacy risks in simple language so you know what’s happening and how to protect yourself.
© 2026 Securityish
