EN
Sign inGet started
Glossary

Token Smuggling

What is Token Smuggling?

Token smuggling is a form of jailbreaking in which unwanted instructions are disguised, for instance split into fragments, re-encoded or written with look-alike characters, so that simple keyword filters miss them while the language model still grasps their meaning. Protection includes normalising inputs before they are assessed, content-based rather than purely word-based checks, reviewing outputs and granting AI agents as few permissions as possible.

How KI.ordo helps

The KI.ordo template “AI firewall: check input” has suspicious inputs assessed for their content before they reach an agent and blocks them if necessary.

Learn more →: Token Smuggling

Matching templates

Topics

  • Security
  • Jailbreaking
  • LLM

Try KI.ordo

Agents, flows and templates in one workspace – free 24-hour demo.

Start demo
← All terms

Related terms

Jailbreaking

Jailbreaking refers to attempts to bypass an AI model's safety rules through cleverly worded inputs, so that it produces content or takes actions it is meant to refuse. For organisations this is a risk, because a chatbot could, for example, reveal confidential instructions or respond inappropriately. Protective measures include clear system instructions, input and output screening, limited permissions, logging and regular security testing (red teaming).

  • Security
  • Misuse prevention
  • +1
Template

AI Firewall

An AI firewall is a protective layer between users, data sources and an AI model. It screens inputs for manipulation attempts such as prompt injection or jailbreaks and checks outputs, for example for confidential data or unwanted content, before passing them on. It may use rules, classification models or a second language model. An AI firewall reduces risk but does not offer complete protection and replaces neither limited permissions nor testing.

  • Security
  • Prompt injection
  • +1
Template

LLM security

LLM security covers all measures that protect applications built on large language models from misuse, data leakage and unwanted behaviour. Typical risks include prompt injection, disclosure of confidential instructions or data, and agents with excessive permissions. Protection comes from several layers: clear instructions, restricted rights, checks on inputs and outputs, AI firewalls, red teaming and regular audits.

  • security
  • language models
  • +1
Template

Indirect Prompt Injection

In an indirect prompt injection, manipulative instructions are not in the user's input but hidden in content an AI system processes along the way, such as web pages, emails or documents. When an agent reads such content, it may mistake hidden commands for genuine tasks and, for example, disclose data. Safeguards include treating external content strictly as data, limiting permissions, screening inputs and requiring human approval for critical actions.

  • Security
  • Prompt injection
  • +1
Template