1---
2title: AI Improvement and Data Collection - Zed
3description: Zed's opt-in approach to AI data collection for improving the agent panel and edit predictions.
4---
5
6# Zed AI Improvement
7
8## Agent Panel
9
10### Opt-In
11
12When you use the Agent Panel through any of these means:
13
14- [Zed's hosted models](./subscription.md)
15- [connecting a non-Zed AI service via API key](./llm-providers.md)
16- using an [external agent](./external-agents.md)
17
18Zed does not persistently store user content or use user content to evaluate and/or improve our AI features, unless it is explicitly shared with Zed. Each share is opt-in, and sharing once will not cause future content or data to be shared again.
19
20> Note that rating responses will send your data related to that response to Zed's servers.
21> **_If you don't want data persisted on Zed's servers, don't rate_**. We will not collect data for improving our Agentic offering without you explicitly rating responses.
22
23When using upstream services through Zed's hosted models, we require assurances from our service providers that your user content won't be used for training models.
24
25| Provider | No Training Guarantee | Zero-Data Retention (ZDR) |
26| --------- | ------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------- |
27| Anthropic | [Yes](https://www.anthropic.com/legal/commercial-terms) | [Yes](https://privacy.anthropic.com/en/articles/8956058-i-have-a-zero-data-retention-agreement-with-anthropic-what-products-does-it-apply-to) |
28| Google | [Yes](https://cloud.google.com/terms/service-terms) | [Yes](https://cloud.google.com/terms/service-terms), see Service Terms sections 17 and 19h |
29| OpenAI | [Yes](https://openai.com/enterprise-privacy/) | [Yes](https://platform.openai.com/docs/guides/your-data) |
30
31When you use your own API keys or external agents, **Zed does not have control over how your data is used by that service provider.**
32You should reference your agreement with each service provider to understand what terms and conditions apply.
33
34### Data we collect
35
36For prompts you have explicitly shared with us, Zed may store copies of those prompts and other data about the specific use of the Agent Panel.
37
38This data includes:
39
40- The prompt given to the Agent
41- Any commentary you include
42- Product telemetry about the agentic thread
43- Metadata about your Zed installation
44
45### Data Handling
46
47Collected data is stored in Snowflake, a private database. We periodically review this data to refine the agent's system prompt and tool use. All data is anonymized and stripped of sensitive information (access tokens, user IDs, email addresses).
48
49## Edit Predictions
50
51By default, when using Zed Edit Predictions, Zed does not persistently store user content or use user content for training of its models.
52
53### Opt-in
54
55Users who are working on open source licensed projects may optionally opt-in to providing model improvement feedback. This opt-in occurs on a per-project basis. If you work on multiple open source projects and wish to provide model improvement feedback you will have to opt-in for each individual project.
56
57When working on other projects where you haven't opted-in, Zed will not persistently store user content or use user content for training of its models.
58
59You can see exactly how Zed detects open source licenses in: [license_detection.rs](https://github.com/zed-industries/zed/blob/main/crates/edit_prediction/src/license_detection.rs).
60
61### Exclusions
62
63Zed will intentionally exclude certain files from Predictive Edits entirely, even when you have opted-in to model improvement feedback.
64
65You can inspect this exclusion list by opening `zed: open default settings` from the command palette:
66
67```json [settings]
68{
69 "edit_predictions": {
70 // A list of globs representing files that edit predictions should be disabled for.
71 // There's a sensible default list of globs already included.
72 // Any addition to this list will be merged with the default list.
73 "disabled_globs": [
74 "**/.env*",
75 "**/*.pem",
76 "**/*.key",
77 "**/*.cert",
78 "**/*.crt",
79 "**/secrets.yml"
80 ]
81 }
82}
83```
84
85Users may explicitly exclude additional paths and/or file extensions by adding them to [`edit_predictions.disabled_globs`](https://zed.dev/docs/reference/all-settings#edit-predictions) in their Zed settings.json:
86
87```json [settings]
88{
89 "edit_predictions": {
90 "disabled_globs": ["secret_dir/*", "**/*.log"]
91 }
92}
93```
94
95### Data we collect
96
97For open source projects where you have opted-in, Zed may store copies of requests and responses to the Zed AI Prediction service.
98
99This data includes:
100
101- sampled edit prediction examples (cursor context + recent diffs/edits) for offline evaluation
102- the edit prediction
103- a portion of the buffer content around the cursor
104- a few recent edits
105- the current buffer outline
106- diagnostics (errors, warnings, etc) from language servers
107
108### Data Handling
109
110Collected data is stored in Snowflake, a private database. We periodically select training samples from this data. All data is anonymized and stripped of sensitive information (access tokens, user IDs, email addresses). The training dataset is publicly available at [huggingface.co/datasets/zed-industries/zeta](https://huggingface.co/datasets/zed-industries/zeta).
111
112### Model Output
113
114We then use this training dataset to fine-tune [Qwen2.5-Coder-7B](https://huggingface.co/Qwen/Qwen2.5-Coder-7B) and make the resulting model available at [huggingface.co/zed-industries/zeta](https://huggingface.co/zed-industries/zeta).
115
116## Applicable terms
117
118Please see the [Zed Terms of Service](https://zed.dev/terms-of-service) for more.