Shadow AI risk in banks: Kaspersky issues warning

A global study by Kaspersky has revealed that 'shadow AI' tools used by employees in the banking, finance, and insurance sectors without institutional approval pose a data security risk. 37 percent of participants in the financial sector consider this a widespread risky behavior.

Kaspersky shared the results of its global research regarding artificial intelligence usage and security threats in the banking, financial services, and insurance sector. According to the research, financial institutions use artificial intelligence at a rate of 59 percent for data analysis, while this rate stands at 54 percent across the general market. In project management, the usage rate for financial institutions was recorded at 54 percent, with a market average of 49 percent. Additionally, 19 percent of financial institutions plan to increase their use of external AI tools, and 8 percent plan to increase internal AI tools.

The use of third-party artificial intelligence tools such as ChatGPT and DeepSeek by employees without permission from their companies creates the 'shadow AI' risk. While 37 percent of participants in the financial sector view this situation as a widespread risky behavior, the general market average remains at 30 percent. The sharing of customer information, financial records, and corporate documents with unapproved tools can cause data to move outside traditional security systems. While 85 percent of participants in the financial sector state that they are aware of the risks, 9 percent believe these risks cannot be completely eliminated.

The research showed that financial companies act more cautiously when using artificial intelligence in software development processes. While 23 percent of organizations in the sector use artificial intelligence in programming, the market average was 34 percent. Kaspersky Artificial Intelligence Technologies Research Center Group Manager Vladislav Tushkanov emphasized that financial institutions must establish clear rules regarding artificial intelligence usage and determine policies to prevent the spread of shadow AI.