Your AI Project Went Live. Who Is Running It Now?

After an AI system goes live, the project team moves on to its next priority. Some teams have quietly returned to their old processes, while outputs that once seemed consistent have become harder to trust. Someone raised a concern informally, but no action followed because no one knew who was responsible for investigating it and following up.

This is an accountability gap. The organisation has no clear owner responsible for assessing the problem, deciding what needs to happen, and ensuring that action is taken.

What Happens When Ownership Is Left Undefined

During implementation, responsibilities tend to be organised around configuring the system, testing it, and completing the rollout. After launch, the organisation still needs someone to check whether the system continues to work as intended.

As usage patterns, workflows, and business needs change, someone must identify the signals, assess their impact, and decide how to respond. Without a defined owner, that follow-up can fall between the project team, technical support, and business users.

The NIST AI Risk Management Framework addresses this through its MANAGE function, which includes post-deployment monitoring, user feedback, incident response, change management, and decommissioning when necessary. Its guidance treats AI risk management as an ongoing responsibility across the system's lifecycle. For organisations, the practical implication is post-launch monitoring needs an owner who can act on what it reveals.

Accountability Means More Than Monitoring

To keep the system working as intended, the organisation needs clear ownership for the decisions and follow-up that monitoring can trigger. That responsibility should cover four areas:

• Performance: Who is responsible for tracking whether the system is still delivering the results it was built for, and for stepping in when it isn’t?

• Human oversight: Who reviews outputs, handles exceptions, and makes decisions that require human judgement?

• Issue management: When something goes wrong, who investigates problems, escalates risks, and who has the authority to take corrective action?

• Continuous improvement: Who collects feedback from the people using the system and decides when it needs to be adjusted, expanded, or retired?

Each area needs a named owner and a clear escalation path. Otherwise, concerns can be identified without anyone being responsible for resolving them.

Establish Ownership Before Launch

Leaders can put this arrangement in place by making three decisions before the system goes live:

• Name an accountable business owner.

  Assign a specific person responsible for the system's ongoing performance and business value. Technical maintenance may sit with IT or an external supplier, but business accountability must remain clear.

• Set the review process.

  Agree on what will be measured, how often performance will be reviewed, and where employees can report issues or share feedback.

• Define decision-making authority.

  Clarify who can approve changes, escalate risks, restrict use, or pause the system when necessary.

These decisions give the organisation a practical way to monitor performance, respond to concerns, and improve the system as needs change.

Before launch, leaders should be able to name the person accountable for the system's ongoing performance, explain what they are responsible for, and clarify the decisions they can make. If those answers are unclear, the organisation has deployed the AI system without establishing who is running it.