Exploring prompt injection defences with a local LLM bytewhispersecurity.com 1 points by quacksilver 5 hours ago