I Couldn’t Trick the AI. So I Hid My Instructions in Its Data.
中文摘要
AI安全研究人员通过操纵训练数据来测试AI漏洞。
English Summary
AI researcher tested LLM vulnerabilities by manipulating training data.
Original Excerpt
Red-teaming LLM and agentic AI from the inside: what AI builders and developers miss about GenAI security and governance Continue reading on Medium »