Who is accountable when your bank's AI says no?

Picture the letter.

Your loan was declined. The reason, in the language regulators now warn against, is that the model said so.

That letter is illegal. And it's been illegal for fifty years.

What's the pattern?

In 1974, the United States passed the Equal Credit Opportunity Act. It was written for a world where a loan officer could look at a woman and decide she needed her husband's signature. Among other things, it required lenders to give a specific reason when they said no. Not "insufficient credit." The actual factors.

That reason requirement, the adverse action notice, is the part that matters now.

Because the model can absolutely say no. It can say it faster and more consistently than any loan officer. What it often can't do is say why, in a way that means anything to the applicant. And in 2023, the Consumer Financial Protection Bureau made it explicit: a complex algorithm is not an excuse. If you use it, you still owe the reason.

Every era of automation in finance has run into this. The credit score itself faced it in the 1990s. Regulators didn't ban the score. They required that it could be explained. The same will happen here. The banks that build the explanation in early will move fast. The ones that bolt it on later will pay for it in enforcement.

What should leaders do now?

So, three things.

Make "explain the no" a design requirement, not a compliance afterthought. If the model can't produce a reason a person could act on, it isn't finished.

Name the human. For every automated decision that affects a customer, one person should be accountable for the outcome. Not a committee. A name.

And test for the answer you'd be embarrassed by. Run your declines through one question: would we be comfortable reading this reason aloud to the customer? If not, the model isn't the problem. The silence is.

Accountability didn't get harder when the machine arrived. The rule was always the same. Say why.

Could your bank explain its last hundred declines in one sentence each?

Related: Would you let an AI agent spend your money?

Sam Rad, The Change Futurist. Keynote speaker on change, transformation, resilience, and AI adoption. Author of Radical Next. Book a keynote