Hackers Can Bypass OpenAI Guardrails Framework Using a Simple Prompt Injection Technique Cyber Security News