Banks have an important opportunity to lead on AI safety

Fotolia

The Trump administration’s repeal of the Biden administration’s AI safety order might sound like a green light for unregulated innovation. But having fewer government rules doesn’t mean banks are free from responsibility. In fact, it may be a rare moment for the industry to set its own high standards for using artificial intelligence — especially now that big decisions about fairness and transparency are banks’ to make.

Financial services already rely on AI for crucial tasks, from approving loans to detecting fraud. A well-tuned algorithm can slash operating costs and boost customer satisfaction. Yet a risky model, left unchecked, can discriminate against certain groups or make baffling decisions that erode trust. Biden’s order, while not perfect, offered a basic framework around model transparency and safety checks. Removing that framework places the job of self-regulation back in the hands of bankers and compliance teams.

Why might this be a good thing? First, banks get to shape guidelines that move at the same pace as their evolving AI projects. Federal rules often lag behind tech changes, forcing bankers to navigate outdated mandates or wait for rule updates. If a bank runs AI for fraud detection, for example, being stuck with slow-moving, one-size-fits-all guidance can hold it back. Now, the bank can tailor more flexible policies that evolve quickly with each new data source or model upgrade.

Second, the industry setting its own bar for AI safety could boost public trust. The public might think banks would jump at a chance to cut corners, but modern banking thrives on reputation. A single AI fiasco — where a large group of customers are wrongly denied accounts or slapped with unfair fees — could spark a national uproar. By publicizing that you test AI models for fairness, bias and accuracy, a bank can position itself as a leader in consumer protection. This matters more than ever, since customers increasingly expect personalized services (like dynamic loan offers) without surrendering their data privacy or dealing with arbitrary decisions.

Third, self-regulation doesn’t have to be lonely or chaotic. Banks can band together — perhaps under an industry association — to create shared best practices for AI in financial services. Think about how payment networks or anti-money-laundering initiatives often involve cooperation across the sector. If multiple banks adopt similar principles around interpretability and data governance, it reassures both regulators and the public that everyone is taking AI risk seriously. In turn, that unity might reduce the chance of future heavy-handed mandates, because lawmakers could see a functioning, transparent ecosystem of bank-led AI oversight.

Even without a…