Anthropic’s Best-of-N jailbreak technique proves how introducing random characters in a prompt is often enough to successfully bypass AI restrictions. https://decrypt.co/298397/anthropic-jailbreak-bypass-ai-restrictions-censorship