Reverse CAPTCHA: Evaluating LLM Susceptibility to Invisible Unicode Instruction Injection

https://www.reddit.com/r/netsec/comments/1rfjlyh/reverse_captcha_evaluating_llm_susceptibility_to/