Google warns its 1.8 billion Gmail users of a new cybersecurity threat involving indirect prompt injections that exploit AI advancements. The company is enhancing protections in its AI chatbot, Gemini, to prevent unauthorised access to sensitive information.