Artificial Intelligence
What AI and large language models really do, where they fail, and the new security risks they bring.
4 explanations, easiest first
What is indirect prompt injection?
Indirect prompt injection is text hidden in a web page that gives orders to your AI (artificial intelligence) agent. If the agent obeys, it acts with your permissions.
How does an AI like ChatGPT really work?
An LLM (large language model) like the one behind ChatGPT doesn't pull its answer from a database. It splits your text into tokens and predicts the likely next part, one token at a time.
What's the difference between an LLM, RAG and an AI agent?
An LLM (large language model) predicts text, RAG (retrieval-augmented generation) pastes looked-up pieces into its prompt, and an AI agent is a loop where the model picks each step and the software around it acts.
How does ChatGPT actually write its answers?
ChatGPT writes its answer one small piece at a time. It predicts a likely next token, adds it and repeats, and nothing in that loop checks whether the result is true.