Most firms want autonomous IT but refuse to let AI act without human approval
While companies aim for autonomous operations, the majority insist on human oversight for AI-driven operational decisions, creating a significant implementation gap.
Key points
- Organizations plan to shift toward autonomous IT operations within the next two years.
- The majority of companies hesitate to allow AI systems to make operational decisions without human approval.
- Few organizations have the foundational groundwork required to support even limited autonomous workflows.
A significant gap exists between organizational goals and current operational realities regarding artificial intelligence in IT management. According to a report by Help Net Security, companies are aiming to move toward autonomous IT operations over the next two years, yet the majority remain unwilling to allow AI systems to execute decisions without human intervention.
The data indicates that while there is strong interest in automating complex tasks, trust in fully autonomous systems remains low. Seventy-seven percent of respondents stated that their organizations hesitate to let AI make an operational decision without a human approving it. This hesitation persists despite the broader industry trend toward adopting agentic AI, a category of software designed to plan and carry out multi-step tasks independently.
The Reality of Autonomous IT
The concept of autonomous IT relies heavily on agentic AI, which differs from traditional automated scripts by its ability to reason and execute multi-step workflows. However, the adoption of these technologies is being slowed by risk aversion. Most companies that express interest in autonomy actually prefer a model that includes a person at the approval step, rather than true end-to-end automation.
This preference for human-in-the-loop systems highlights a cautious approach to integrating AI into critical infrastructure. Organizations are recognizing the potential efficiency gains of autonomous operations but are unwilling to cede control entirely to algorithmic decision-making. Consequently, the definition of "autonomy" in these environments often centers on assistance and recommendation rather than independent action.
Implementation Challenges
Beyond the cultural hesitation to remove human oversight, technical readiness poses a barrier. The report from Help Net Security notes that few organizations have the necessary groundwork in place to support autonomous operations, even at a limited level. Establishing the data integrity, system connectivity, and security protocols required for agentic AI to function safely is a complex undertaking.
This lack of foundational infrastructure means that many companies are not yet prepared to scale AI-driven operations, regardless of their strategic goals. The transition from manual or semi-automated IT management to a state where AI can independently execute tasks requires significant preparation. Without this groundwork, attempts to deploy autonomous systems may lead to errors or security vulnerabilities, reinforcing the reluctance to proceed without human checks.
What should you do?
Organizations looking to advance toward autonomous IT operations should focus on assessing their current infrastructure readiness. Before deploying agentic AI, ensure that data pipelines are clean and that system APIs are secure. Implement rigorous logging and monitoring to track AI actions during pilot phases. Maintain clear protocols for human intervention and approval workflows to manage risk while testing automated capabilities.
What to do and how to stay safe: agentic AI
- Audit existing IT infrastructure to ensure it has the necessary data integrity and connectivity for agentic AI tools.
- Implement robust logging and monitoring systems to track all actions taken by AI agents during testing phases.
- Define clear human-in-the-loop approval workflows for critical operational decisions before attempting full automation.
- Conduct risk assessments to identify potential security vulnerabilities introduced by autonomous decision-making processes.
Step-by-step guide: Prompt Injection Warning Signs: How to Spot AI Manipulation
General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
What percentage of companies want to move toward autonomous IT operations?
According to the report, ninety percent of companies want to move toward autonomous IT operations over the next two years.
How many companies are willing to let AI make decisions without human approval?
Seventy-seven percent of companies say their organization hesitates to let AI make an operational decision without a human approving it.
What is agentic AI?
Agentic AI is software that plans and carries out multi-step tasks on its own.
