Malicious web prompts can weaponize AI without your input. Indirect prompt injection is now a top LLM security risk. Don't treat AI chatbots as fully secure or all-knowing. Artificial intelligence (AI ...