Disclaimer: This article doesn’t aim to be an exhaustive study but rather a collection of interesting findings. — TL;DR: ChatGPT appears to effectively handle negations in most cases. Negations are an effective way to circumvent ChatGPT’s security mechanisms. Language Models (LM) are known for their shortcomings when handling negations, as shown in (Ettinger, 2019). But, does this also hold in the case of ChatGPT? Let’s take a look…