A Prompt Injection Vulnerability occurs when user prompts alter the LLM's behavior or output in unintended ways. These inputs can affect the model even if they are imperceptible to humans.

Sensitive information can affect both the LLM and its application context, including PII, financial details, health records, confidential business data, security credentials and legal documents, as well as proprietary model details.

LLM supply chains are susceptible to various vulnerabilities, which can affect the integrity of training data, models and deployment platforms, resulting in biased outputs, security breaches or system failures.

Data poisoning occurs when pre-training, fine-tuning or embedding data is manipulated to introduce vulnerabilities, backdoors or biases, compromising model security, performance or ethical behavior.

Improper Output Handling refers specifically to insufficient validation, sanitization and handling of the outputs generated by large language models before they are passed downstream to other components and systems.

Excessive Agency is the vulnerability that enables damaging actions in response to unexpected, ambiguous or manipulated outputs from an LLM granted the ability to call functions or interface with other systems via extensions (tools, skills or plugins). Root causes include excessive functionality, excessive permissions or excessive autonomy.

The system prompt leakage vulnerability refers to the risk that the system prompts or instructions used to steer the behavior of the model can also contain sensitive information that was not intended to be discovered.

Vectors and embeddings vulnerabilities present significant security risks in systems utilizing Retrieval Augmented Generation (RAG) with LLMs: weaknesses in how vectors and embeddings are generated, stored or retrieved can be exploited to inject harmful content, manipulate outputs or access sensitive information.

Misinformation occurs when LLMs produce false or misleading information that appears credible, often caused by hallucination, and can lead to security breaches, reputational damage and legal liability; overreliance compounds the harm.

Unbounded Consumption covers attacks designed to disrupt service, deplete the target's financial resources (denial of wallet), or even steal intellectual property by cloning a model's behavior, by exploiting uncontrolled inference.