Hackers can exploit nine popular AI tools to create massive botnets

Prompt injection has become the top threat in AI security because large language models cannot distinguish between legitimate user instructions and malicious commands hidden in third-party content. With no way to enforce boundaries between trusted and untrusted sources, developers rely on guardrails rather than solving the root cause. To date, most prompt injections have been 'push' attacks that target each victim individually, which limits their scalability for mass exploitation.