A researcher has successfully exploited systemic vulnerabilities in large language models (LLMs) to bypass safety mechanisms and obtain hazardous instructions, exposing a widespread security issue across the industry. The discovery, made by Dave Kuszmar, affects nearly all major LLMs, highlighting the need for increased transparency and rigorous safety research. Kuszmar's findings suggest that the current pace of LLM deployment is outstripping the industry's ability to ensure their safe operation, and he advocates for a more cautious approach to integration1. The implications of this research are far-reaching, as LLMs are increasingly being embedded in various aspects of society. This vulnerability has significant consequences for the security and reliability of these systems, making it essential to address these concerns before further deployment. The fact that these exploits can be applied across multiple LLMs underscores the urgency of this issue, so what matters most to practitioners is the need to prioritize LLM safety and security to prevent potential misuse.