AI Agents: What Should Managers Understand Before Using Them?
Before using an AI agent, understand what it can access, which actions need approval and who reviews the outcome.
About This Update
For: Managers, Operations, Learning, IT and data-protection teams.
Source published: 8 October 2026.
Status: Regulatory update and call for evidence. This is not new legislation or final guidance.
Confirmed deadline: Responses close at the end of 20 November 2026. This is a consultation deadline, not a deadline for organisations to implement new controls.
From Producing Answers to Taking Actions
AI tools are increasingly able to complete tasks, use other tools and interact with websites. That makes it useful to look beyond the quality of an answer and understand what the system can actually do.
The ICO has opened a call for evidence on the data-protection risks of AI agents. It is seeking information about security, transparency, accountability, automated decisions, fairness and lawful data use. The evidence will help inform its future guidance.
My Practical Perspective
I believe learning should connect the technology to the work people are doing. Before introducing an AI agent, a useful team discussion would explore:
- What task are we asking it to complete?
- What information and systems can it access?
- Which actions need a person’s approval?
- Who checks the outcome and handles an unexpected result?
- How can someone stop the process and raise a concern?
For example, drafting a customer response and sending that response are different activities. Managers and colleagues need to understand where their responsibility sits and which actions are permitted.
Start with One Use in Your Organisation
A practical starting point is to map one proposed use with the people responsible for the process, technology and data protection. Agree the boundaries, practise relevant situations and review what happens during an approved pilot.
These are my suggestions for applying the update. They are not a list of new ICO requirements or a substitute for the organisation’s data-protection assessment.
I can help organisations turn those agreed arrangements into clear learning, manager guidance and process support, then evaluate how well people apply them.
Make AI Learning Relevant to the Work
I can help you connect agreed AI boundaries with practical learning, manager support and process improvement.
Start a conversation