Google has issued a warning about a new cybersecurity threat involving AI-powered prompt injections, where hackers manipulate AI systems to steal user information, particularly targeting its AI assistant, Gemini. This technique, termed “indirect prompt injection,” allows attackers to embed harmful prompts in benign content, deceiving AI into acting against its own protocols. As more users rely on AI for personal tasks, the risks of such sophisticated scams are increasing. Google is enhancing protective measures and advising users to remain vigilant against these evolving cyber threats that blend AI capabilities with malicious intent.