Your AI ROI Might Look Great. Have You Calculated the Liability?

Businesses are measuring what artificial intelligence saves them. But are they measuring what it could cost them?

The business case for artificial intelligence has become remarkably predictable.

Reduce operating costs. Increase productivity. Automate repetitive tasks. Save employees time. Improve efficiency.

These are legitimate business objectives. Organizations should be looking for ways to operate more effectively.

But there is a question missing from far too many conversations about AI investment:

What happens to your projected return when you account for the financial consequences of AI failure?

Calculating AI ROI requires leaders to look beyond projected savings and productivity gains. A more complete assessment considers implementation and oversight costs, potential financial losses, insurance limitations, and whether the organization is prepared to manage AI-influenced decisions. These factors help leadership evaluate whether the expected benefits justify the investment and the risks being accepted.

Consider a business that deploys an AI tool to improve customer service, process claims, evaluate financial information, or recommend actions to employees.

The projections are impressive. The technology promises thousands of hours in savings and significant operational efficiencies. Leadership approves the investment.

Six months later, the organization discovers that the tool has been producing inaccurate recommendations. Employees relied on those recommendations, customers were harmed, and the organization now faces remediation costs, potential regulatory scrutiny, and litigation exposure.

What happened to the projected savings? More importantly, was that possibility ever considered when leadership approved the investment?

The Hidden Costs Missing From AI ROI Calculations

An AI investment may deliver genuine operational savings while simultaneously introducing financial exposure. That exposure can take several forms:

Legal and regulatory costs: Investigations, litigation, defense expenses, settlements, and potential penalties.

Data and security incidents: Breach response, notification expenses, business interruption, and recovery costs.

Customer and client harm: Compensation, remediation, lost business, and damaged relationships.

Operational failures: Incorrect recommendations, unreliable outputs, costly rework, or decisions requiring reversal.

Oversight and control costs: Testing, monitoring, employee training, documentation, and human review.

Some of these costs are reasonably predictable. Others depend on the likelihood and severity of events that may never occur.

But uncertainty is no justification for leaving them out of the decision.

A responsible financial assessment should evaluate plausible loss scenarios, estimate exposure where credible data exists, and identify significant risks that cannot be reliably quantified.

Otherwise, leadership may be approving an investment based on a return that looks considerably better on paper than it does when the risks are considered.

AI Liability and Insurance Coverage Gaps

There is another development business leaders should be watching closely. The insurance industry is reconsidering how it covers AI-related losses.

In January 2026, Verisk’s Insurance Services Office made optional generative AI exclusion endorsements available for commercial general liability policies. These endorsements can restrict coverage for certain claims involving bodily injury, property damage, and personal or advertising injury arising from generative AI.

Some insurers are introducing their own AI-related exclusions into other forms of liability coverage. This does not mean every AI-related claim is excluded or that all policies have the same restrictions.

It means businesses can no longer safely assume that existing insurance arrangements will automatically respond when AI contributes to a loss. An organization may discover that it has retained financial exposure it expected to transfer to its insurer.

If your AI investment creates a risk your insurance will not cover, that exposure belongs in your investment decision.

What should leadership be asking?

1. What business problem are we solving?

Define the desired outcome, measurable benefits, and alternatives to deploying AI.

2. What could go wrong, and who could be harmed?

Consider inaccurate outputs, improper reliance, security failures, and consequential decisions affecting customers, employees, or third parties.

3. What would those failures cost?

Develop reasonable financial scenarios where possible. Identify costs of prevention, detection, correction, and recovery. Make material uncertainty visible.

4. Who owns the decision and the risk?

Establish who approves deployment, monitors performance, reviews consequential outputs, authorizes intervention, and can pause the system.

5. What financial exposure are we actually retaining?

Have appropriate professionals review insurance policies, contractual allocations, indemnities, exclusions, and coverage limits.

The return must justify the risk

Every investment carries some degree of uncertainty. AI is no exception. The goal is not to eliminate every conceivable risk or demand financial precision where none is possible. The goal is to make informed decisions.

A projected return becomes less persuasive when leadership cannot explain its assumptions, the exposure accepted, or the safeguards needed to protect customers and the business.

The next time someone presents an impressive ROI for an AI investment, ask one more question: What happens to that return when something goes wrong?

At Equity Principle Consulting, we help leaders assess where AI belongs, prepare their organizations for change, and establish accountability for AI-influenced decisions. That work begins with understanding what the organization stands to gain, what it could lose, and who is responsible for the decisions made along the way.

Considering an AI investment? Let’s talk before you commit.