Banking institutions, including NatWest and Bank of America, issued a warning on Tuesday (22) regarding the dangers associated with the use of Artificial Intelligence (AI) agents in online purchases. These risks encompass scams, fraud, and privacy violations, according to Reuters.
Currently, AI technology is being used to assist consumers in product research, item selection, and transaction completion. However, banks point out that the development of this technology is outpacing the creation of adequate security standards and defense mechanisms.
Major companies such as OpenAI, Anthropic, Google, and Meta are promoting chatbots as purchasing tools. The idea is to integrate AI agents into phases that traditionally required direct consumer intervention, such as choosing merchandise and finalizing orders.
The retail sector is also looking to capitalize on this new landscape, influencing the suggestions offered by the systems. For example, in September, the British store John Lewis reported that searches originating from AI agents represented 2.5% of its traffic, a significant increase compared to the 0.3% recorded a year earlier.
Financial institutions have consolidated these concerns in a document focused on what they call agent-mediated commerce. While the material acknowledges public interest in this technology, it warns that existing protections and security standards have not yet kept pace with the rapid pace of these systems.
Detailed risks in the document
The report details scenarios that could put consumers at a loss. One point raised is the decision-making capacity of the AI agents themselves, as the document states that 'consumers are not entirely sure if the AI will act in their interests.'
Another serious concern relates to access to card data. An AI agent can request this information and enter it directly onto a website. There is also the risk that the system might divert the customer to payment methods with a lower level of protection.
Among the problems highlighted by the banks is the uncertainty about what should happen if a purchase made by an agent results in a failure. Consumers are still unclear whether they will be legally protected or whom they should turn to.
The banks plan to discuss proposals with regulators and policymakers. One such suggestion is to require that the consumer be notified whenever an AI agent is involved in a transaction. Furthermore, the group advocates for greater transparency regarding how systems reach their decisions and the implementation of specific mechanisms to safeguard customer data.
The report emphasizes the fear that AI agents might acquire incorrect items or overspend, or, in more serious cases, lead the user into scams and fraud.
The institutions also argue that both consumers and merchants should have the freedom to select the AI-based commerce services they wish to use. Another important aspect is the interoperability between systems, allowing various technologies to function together. The proposals focus the discussion on crucial topics such as transparency, data protection, and defining responsibility for transactions executed by AI agents.