Vyro AI, a maker of AI-content creation applications, has inadvertently leaked 116GB of sensitive user data from three of its products: ImagineArt, Chatly, and Chatbotx. The leaked data includes AI prompts that users type into the applications, bearer authentication tokens, and user agents. The company, based in Pakistan, has more than 150 million app downloads across its portfolio, according to Cybernews researchers. And considering that some individuals are inputting sensitive information during their conversations with AI, breaches such as these could become particularly revealing, they added. If resources such as ChatGPT, Gemini, and Copilot are the only options available, he adds, companies should implement policies and guides to restrict sharing sensitive information with the AI model.