Concern 435 · Working draft · editorial review in progress
Training-data disclosure
Training-data disclosure occurs when an AI model reproduces private information it encountered while learning.
Understanding the concern
Training changes a model's internal numerical settings. Some information can be memorised closely enough to appear later in an output. Researchers have demonstrated this in particular models and conditions. It differs from a live database breach because the exposed text may come from learned parameters rather than a retrieved account record.
Think of someone unexpectedly repeating a private passage they once read. Removing the original document from a website does not necessarily remove its influence from an already trained model. Reducing unnecessary personal data in training and testing for disclosure can help. A model is not a perfect searchable copy of all training material, however.
An example
A generated answer unexpectedly includes a private passage that appeared in the model's training data.
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.
- Database
- An organized collection of information that software can store, search, and update.
Evidence and its limits
The research demonstrates extraction under stated conditions; it does not mean every training record can be recovered from any model.
Carlini and colleagues · USENIX Security 2021 · Research study
Extracting Training Data from Large Language Models ↗ (opens in a new tab)Demonstrates training-text recovery in tested models, including public identifying details; not every output resembling personal data is a verified leak.
European Data Protection Board · December 2024 · Regulatory opinion overview
Opinion on personal data and AI models ↗ (opens in a new tab)Discusses anonymity, legitimate interest, and unlawful processing case by case. Consent is not the only potential lawful basis.
Broader frameworks and background
These resources provide context. They do not establish every claim in this concern.
- OECD AI Principles (opens in a new tab)
framework
- NIST AI RMF (opens in a new tab)
framework
A question to discuss
What personal data was included in training, and how is the model tested for unwanted reproduction?
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.
Dignity
Dignity means recognizing each person's worth regardless of usefulness, status, or a system's score. People deserve respect and a voice in matters that affect them.
Explain important decisions, avoid humiliating treatment, and keep room for a person's story. Do not treat a prediction as a complete account of who someone is.
Ask together: Does this use treat each person as someone whose experience matters?
Trustworthiness
Trustworthiness means giving people good reasons to rely on you. It grows when promises, actions, and evidence line up. With AI, a friendly voice or a famous brand is not enough: people need to know what was checked and what could still go wrong.
In practice, keep promises small enough to test, record mistakes, and tell people when the tool's limits affect them. If a service promises human review, make sure a person really has time and authority to review it.
Ask together: What would give the people affected a sound reason to trust this use?
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?
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.