# Limited-Risk AI

> AI that may be used freely, as long as people are told it is AI.

**Field:** Artificial Intelligence  
**Category:** Ethics & Regulation  
**Also known as:** Transparency Risk (EU AI Act), Limited Risk (EU AI Act), AI Act Transparency Obligations

## Think of it like

It is like a sign on a shop door that says a robot works at the till. You may go in, but you are told first.

## Definition

Limited-risk AI is the EU AI Act's level for AI that is fine to use, as long as people are told. The rules are in Article 50 and apply from 2 August 2026. A chatbot must say it is an AI, unless that is plain to see. AI that makes pictures, sound, video or text must mark its output in a way a machine can read. A deepfake must be labelled as fake by whoever puts it out. People must also be told when a system reads their feelings or sorts them by their body.

## Example

A bank in Madrid puts a chatbot on its website. Its first line says, "I am an AI assistant." Its image tool also hides a mark in each picture it makes.

## Why it matters

People trust a voice, a face or a reply more when they think a person made it. Telling them lets them decide how far to trust it.

## Common mistake

Thinking a label makes any use fine. If the same chatbot screens job seekers, it is high risk, and the label is only one of many duties.

## Three principles

- **Tell people.** Say it is AI, unless it is obvious.
- **Mark the output.** Made pictures and sound carry a hidden tag.
- **From August 2026.** The rules apply from 2 August 2026.

## Related terms

- [EU AI Act](https://learn.benibix.com/term/ai/eu-ai-act)
- [AI Disclosure](https://learn.benibix.com/term/ai/ai-disclosure)
- [Deepfake](https://learn.benibix.com/term/ai/deepfake)
- [Content Credentials (C2PA)](https://learn.benibix.com/term/ai/content-credentials)
- [Minimal-Risk AI](https://learn.benibix.com/term/ai/minimal-risk-ai)
- [High-Risk AI System](https://learn.benibix.com/term/ai/high-risk-ai-system)

![A pyramid of four risk levels: banned at the top, then high risk, then limited risk that needs labels, then minimal risk at the wide base.](https://learn.benibix.com/diagrams/et-eu-ai-act.svg)

---

Source: https://learn.benibix.com/term/ai/limited-risk-ai
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-18
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
