Concern 443 · Working draft · editorial review in progress
Federated learning leakage
Federated-learning leakage is the exposure of training information through updates shared by devices or organisations that keep their raw data locally.
Understanding the concern
Federated learning lets participants contribute to a shared model without pooling all their original records. They send updates describing how the model should change. Those updates can still carry information about the data. Research has shown recovery of training examples under some experimental conditions.
Keeping the original file at home is therefore helpful but not a complete privacy guarantee. It is like sending detailed corrections to a document: the corrections may reveal something about the original. Additional protections can include limiting what is shared and using secure aggregation or differential privacy. Their effectiveness depends on the threat model and implementation.
An example
A project promises privacy solely because raw records never leave each participant, without evaluating what model updates reveal.
Illustrative unless identified as a documented case.
Words in plain English
- 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.
- Differential privacy
- A mathematical approach that limits how much a published result reveals about one person's data. The strength depends on how it is set up and used.
- Federated learning
- Training across several devices or organizations while keeping some data in its original location. It does not automatically remove all privacy risks.
Evidence and its limits
The leakage paper demonstrates particular conditions. It does not establish that every federated-learning system exposes all participant data.
Zhu, Liu and Han · Research paper
Deep Leakage from Gradients ↗ (opens in a new tab)Demonstrated recovery from shared training updates under experimental conditions; not universal leakage from every federated system.
Near, Darais and Lefkovitz; NIST · Technical guidance
Guidelines for Evaluating Differential Privacy Guarantees, NIST SP 800-226 ↗ (opens in a new tab)2025 mathematical guarantees, implementation pitfalls and interpretation of privacy loss.
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 can be inferred from the shared updates, and which additional protections have been tested?
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.