Alert Gmail users! Google warns 1.8 billion accounts vulnerable to new AI scam

Google has issued a significant warning to its vast user base of 1.8 billion Gmail accounts about a fresh cybersecurity threat stemming from advancements in artificial intelligence. The alert, reported by Men’s Journal, centers around indirect prompt injections—an insidious form of cyberattack that could potentially impact individuals, businesses, and government entities alike. In a recent blog post, Google elaborated on the nuances of these attacks. Unlike direct prompt injections, where hackers input harmful commands directly into an AI system, indirect attacks cleverly embed malicious instructions within innocuous external sources such as emails, documents, or calendar invites. When these messages are processed, the hidden directives can manipulate the AI into disclosing sensitive information or executing unauthorized actions. The tech giant cautioned that with the increasing integration of generative AI into daily functions, a new wave of threats is surfacing across various sectors. Technology analyst Scott Polderman highlighted that cybercriminals are leveraging Google’s own AI chatbot, Gemini, to perpetrate these scams. He noted that harmful code can be concealed within an email, allowing Gemini to inadvertently extract login credentials without the user’s knowledge. “The real danger is that users don’t need to click on anything,” Polderman explained. “Covert instructions can manipulate the AI into revealing passwords and other sensitive data, effectively turning the system against its users.” In response to these threats, Google has initiated the rollout of enhanced protective measures. These include fortifying its Gemini 2.5 model, deploying machine-learning systems to identify suspicious prompts, and implementing broader security protocols at the system level. The company aims to increase the complexity and costs associated with these attacks, thereby compelling cybercriminals to resort to more detectable and less subtle methods. This warning comes at a time of heightened concern regarding the potential for artificial intelligence to be misused, underscoring the risks associated with integrating AI tools into essential services that billions depend on globally.