This could cost many companies. "The threat is growing"

Experts warn that companies implementing AI chatbots must consider the risk of "Prompt Injection" attacks, where attackers manipulate the system to execute unauthorized commands. This threat is particularly significant for chatbots used in internal communications and customer interactions. If a chatbot has access to sensitive company resources, attackers could potentially delete files or disclose confidential information. The use of personal AI tools in the workplace also poses risks, as employees may inadvertently create insecure applications. To mitigate these risks, companies should secure AI tools at multiple levels, including training staff on potential threats and managing access to sensitive information.