LLM

An LLM is an AI model trained on enormous amounts of text to predict the next word, which turns out to be enough to write, summarize, translate, and answer questions convincingly. Tools like ChatGPT and Claude are built on LLMs. For security teams they cut both ways: they power useful assistants and triage bots, but they also introduce risks like prompt injection and data leakage that traditional application security never had to consider.