Who Is Liable When AI Recommends the Wrong Financial Product? [BKL’s Future Finance]

FSC Says AI Is a Support Tool and Staff Must Intervene

Financial Firms and Employees May Face Greater Liability

Companies Need Rules on the Degree of Human Involvement

Hankyung Law & Biz’s "Law Street" column offers practical legal guidance for companies and individuals. Lawyers cover issues ranging from tax, inheritance, labor, antitrust, M&A and finance to analysis of major court rulings.

Past debates over artificial intelligence often focused on self-driving cars: If an autonomous vehicle caused an accident, was the driver responsible or was the manufacturer? The issue was never limited to cars. It reflected a broader question of whether humans should bear responsibility for machines or systems that make autonomous decisions. Self-driving cars became the center of the debate simply because they were the most visible example.

As AI spreads, a similar dispute has resurfaced. Discussion is growing over who should be held responsible when AI wrongly recommends a financial product to a consumer or when a hiring process is conducted unfairly. In that context, the Financial Services Commission’s revised Guidelines on Artificial Intelligence in the Financial Sector, released this year, offer a useful reference point.

FSC: AI is a support tool, and humans must be involved

The FSC guidelines deal with administrative supervision of financial institutions, so they are not directly tied to civil-law judgments. From a regulatory perspective, however, their direction is clear.

The FSC sets out "support-tool status" as one of seven core principles for the use of AI. The guidelines state that financial companies should use AI as a support tool in their operations, while employees remain responsible for final decision-making and the liability that follows. For high-risk AI in particular, the guidelines say companies need to establish and operate standards that allow internal staff and other human actors to intervene in how the system works.

The principle means AI output should be used as reference material, while human review and judgment continue throughout the workflow. Under that framework, AI is used to support work performed by employees at a financial institution. If a financial incident occurs, that structure could leave the institution bearing primary responsibility.

The guidelines also suggest that responsibility for AI use may rest less with the IT department that led development than with the business unit that actually used the system in its work.

Financial firms and employees may face more liability

That may seem unfair to financial institutions. If firms and employees remain responsible even after adopting AI, some may wonder whether they could introduce such systems without conducting advance testing.

But compliance efforts such as pre-deployment testing can still be reflected when regulators review administrative liability. That makes preventive measures and post-incident controls critical to avoiding accidents tied to AI services.

The FSC guidelines are not regulations enacted under statute. That means the guidelines themselves cannot determine civil liability. Unlike administrative responsibility, detailed discussion of civil damages arising from a financial institution’s use of AI remains underdeveloped.

There is still limited case law and academic debate on who should bear responsibility when flawed AI output causes losses for customers. It remains unclear whether liability would rest with the financial institution, the employee who handled the matter, or an outside vendor that supplied the AI model.

Still, internal rules at individual financial firms are likely to adopt a structure under which employees bear final responsibility for using and verifying AI output. That increases the possibility that employees will be viewed as having been substantively involved in the use of AI. In turn, that could raise the chances that both financial institutions and their employees face civil liability for damages if an incident occurs during AI use.

Rules on the degree of intervention are needed to address liability

Financial services have a strong direct impact on people. Given the current spread and sophistication of AI, the guideline’s principle that AI should serve as a support tool appears necessary in finance. Whether the same principle can be extended to other fields is more open to question.

The FSC’s support-tool principle is worth considering through the lens of liability for AI use. Responsibility for a company or its employees will vary depending on the degree of human involvement. Companies therefore need to set their own principles on that issue and build compliance systems capable of enforcing them. That could provide a clearer way to resolve liability questions arising from the use of AI.

<Hankyung Law & Biz contributors> Yoon Joo-ho, attorney at Bae, Kim & Lee