Concern 155 · Working draft · editorial review in progress
Multilingual jailbreaks
Multilingual jailbreaking exploits differences in how reliably an AI applies safety rules across languages.
Understanding the concern
A model may have much more training and testing in one language than another. Researchers have found cases where a request refused in English receives an unsafe answer after translation. That gap can affect both people seeking help in their language and people deliberately testing safeguards.
Language coverage is therefore a safety issue as well as an access issue. The evidence does not mean that every less-resourced language is unsafe or that its speakers are a risk. Systems need evaluation by people who understand the language and context, not only machine-translated checklists.
An example
A service carefully checks its English entrance but leaves the same rules poorly enforced at another language's entrance.
Illustrative unless identified as a documented case.
Words in plain English
- AI
- Software that learns patterns from examples or uses rules to perform tasks such as recognizing images or generating text. It can make mistakes and does not understand every situation as a person would.
- Model
- The learned part of an AI system. It uses patterns found during training to produce an answer or prediction.
- Training
- The process of using examples to shape how an AI model behaves. Learning patterns from material does not establish that the material is true or fair.
- Jailbreak
- An attempt to get an AI to act outside its intended limits.
Evidence and its limits
The study evaluates a specified GPT-4 version and language set; it is not a current ranking of languages or models.
Yong, Menghini and Bach · 2023 · Research study
Low-Resource Languages Jailbreak GPT-4 ↗ (opens in a new tab)Measures cross-language safety gaps in a specified model and dataset, not inherent risk in a language or its speakers.
Broader frameworks and background
These resources provide context. They do not establish every claim in this concern.
A question to discuss
Which languages have received direct safety evaluation by fluent speakers?
Keep track of different views, what needs checking, and any next step in your own discussion notes.
Open discussion notes →Principles for the conversation
Before you finish · reflect together
Virtues and principles to consider
Virtues are qualities we practice in how we treat people and make choices. They are not scores for judging someone. Use these ideas to explore what a thoughtful response could look like.
Stewardship
Stewardship means caring for resources and systems that others depend on, including people who will use them later. It looks beyond the benefit of a quick launch.
Plan for maintenance, energy use, data protection, and eventual shutdown. Ask who will look after the tool when the original team moves on and how future users can avoid inheriting hidden costs.
Ask together: What are we responsible for caring for over time?
Prudence
Prudence means using careful judgment before and during action. It balances worthwhile opportunities with the need to avoid harm that could have been anticipated.
Start with a limited trial, set stopping points, and prefer choices that can be undone when uncertainty is high. Review what happens rather than assuming the original plan remains sound.
Ask together: What is a careful next step, and when should we stop or change course?
Accountability
Accountability means being answerable for decisions and taking responsibility for repair. It requires more than saying that the software made the choice.
Name the person or team responsible, keep useful records, and give them authority to correct problems. Make the route to a complaint clear to the people affected.
Ask together: Who will answer for the outcome and help put things right?
Editorial ratings
These are editorial judgments from the register, not measured probabilities or predictions about your situation.
About the register and scoring →More ways to explore
Search for more writing
For more specific or recent material, open a search below or ask for a web summary. AI-generated summaries can contain mistakes; open the original articles to check them.