Fasoo AI: "8 in 10 Use 'Shadow AI' Even When Banned at Work"

Nearly 90% of employees at South Korean companies and institutions use generative artificial intelligence (AI) through personal accounts for work, and nearly 8 in 10 continue to use “shadow AI” even when their organizations ban its use outright.

According to Fasoo AI on October 28, a survey of 151 AI, IT and security professionals at major South Korean companies and institutions found that the most common approach to AI use within organizations was to allow employees to use it at their discretion without a separate policy, at 29.1%. This was followed by allowing only designated generative AI services (22.5%) and building and operating in-house AI systems (17.9%). Organizations that said they block AI use in principle accounted for 14.6%, but 88.7% of all respondents said they currently use generative AI through personal accounts for work. Even among respondents whose organizations said they block use in principle, the shadow AI usage rate reached 77.3%.

Only 19.2% of organizations said they monitor and control employees’ use of AI through personal accounts. The top concern about shadow AI was “internal critical data being used to train third-party models” (28.3%). This was followed by “violations of regulations such as the Personal Information Protection Act and GDPR” (24.0%) and “difficulty responding to incidents because usage histories cannot be tracked” (23.7%).

The main reasons employees use shadow AI were “the absence of clear usage policies or guidelines” (28.5%) and “to avoid leaving a usage history on company accounts or systems” (25.3%). “Insufficient in-house AI performance” (19.9%), “the absence of approved AI tools” (14.6%) and “cumbersome application and approval procedures” (11.7%) were also found to contribute to shadow AI use.

As a measure to reduce the risks of shadow AI, respondents most often cited “providing in-house AI tools that employees can use safely” (25.7%). This was followed by “gaining visibility into AI usage” (20.6%), “establishing clear AI usage policies and guidelines” (18.9%) and “technology to control sensitive information at the prompt-entry stage” (17.7%).

Son Jonggon, managing director of Fasoo AI, said, “This survey shows that shadow AI is difficult to prevent through bans alone. The key is to provide clear policies and safe alternatives, and to gain visibility into AI usage.” He added, “Fasoo AI plans to soon unveil an AI security platform that comprehensively manages various data leakage channels and provides visibility into AI usage across the board.”

This content was produced with the assistance of AI translation services.