This overview explains how your data is handled based on your Privacy Mode setting and points to the Privacy Policy
This Data Use & Privacy Overview describes how your data is handled based on the privacy settings you choose in Cursor, and points to the full Privacy Policy for how Anysphere collects, uses, discloses and processes personal data. You can email the stated address with questions or feedback. The page is labeled “Last updated September 3, 2026.” The page does not say whether it forms part of any binding agreement, and the Privacy Policy is a separate document that is not part of this source.
Supporting quotations
how your data is handled based on the privacy settings you choose in Cursor:
If you have any questions or feedback, please email us at hi@cursor.com.
for more information about how we collect, use, disclose, and process your personal data, see our full
Last updated September 3, 2026
Not stated here
Which privacy settings besides Privacy Mode affect data handling is not stated on this page.
Whether this overview has binding contractual effect is not stated.
Use & Processing · Company commitment
With Privacy Mode on, Cursor says Customer Data is not used for training and providers will not store or train on it
If you enable Privacy Mode in Cursor's settings, the page says Customer Data will not be used for training by Cursor. It says Cursor maintains zero data retention (ZDR) agreements with all providers and that AI model providers will not store or train on your data. The same paragraph adds exceptions for risk classifiers and abuse detection and for non-ZDR models, recorded in the next two cards.
Supporting quotations
If you enable
settings: Customer Data will not be used for training by Cursor.
Cursor maintains zero data retention (ZDR) agreements with all providers, and AI model providers will not store or train on your data.
Qualifications & exceptions
Applies when Privacy Mode is enabled; subject to the abuse-detection storage and non-ZDR model statements in the same paragraph.
Not stated here
“Customer Data” is not defined on this page.
Whether the commitment covers every product and plan, and how it relates to the training statements in other Cursor documents, is not stated here.
Retention & Deletion · Restriction or exception
Even with Privacy Mode on, abuse detection may cause your data to be stored for investigation
With Privacy Mode on, the page says model providers, including Cursor, may run risk classifiers, subject to their policies, to detect violations of terms and usage policies. If your prompts or conversations trigger abuse detectors, your data may be stored for investigation and deleted in accordance with their retention policies. This qualifies the statement that providers will not store your data.
Supporting quotations
However, please note that subject to their policies, model providers (including Cursor) may run risk classifiers to detect violations of terms and usage policies, and if your prompts or conversations trigger abuse detectors your data may be stored for investigation and deleted in accordance with their retention policies.
AI model providers will not store or train on your data.
Not stated here
How long data stored for investigation is kept, and who can access it, are determined by the providers' retention policies, which are not supplied here.
Whether data stored under this exception may be used for training is not stated.
Retention & Deletion · Company practice
With Privacy Mode on, non-ZDR models are labeled as such or need an admin to opt in for your workspace
With Privacy Mode on, the page says non-ZDR models will be designated as such or require an admin to opt in to enable the model for your workspace. This indicates some models are not covered by zero data retention agreements.
Supporting quotations
Non-ZDR models will be designated as such or require an admin to opt-in to enable the model for your workspace.
Not stated here
Which models are non-ZDR, and how an individual (non-workspace) user is told, are not stated here.
Linking & Identification · Context
The page links to model providers' documentation for their retention policies (SpaceXAI, OpenAI, Anthropic, Meta)
For more details about the model providers' retention policies, the page points to documentation from SpaceXAI, OpenAI, Anthropic and Meta. That documentation is not part of this source, so this card does not describe any provider's retention practice.
What these providers' retention policies say is not stated in this source.
Use & Processing · Company practice
With Privacy Mode off, Cursor may use and store code data, prompts and editor actions to improve and train its models
If you choose to turn off Privacy Mode, the page says Cursor may use and store codebase data, prompts, editor actions, code snippets and other code data and actions to improve its AI features and train its models. How this relates to the training preferences described in other Cursor documents is not stated on this page.
Supporting quotations
If you choose to turn off
we may use and store codebase data, prompts, editor actions, code snippets, and other code data and actions to improve our AI features and train our models.
Not stated here
The text says “may use and store”; it does not say whether data is used in every case or how long it is kept.
How the Privacy Mode setting interacts with the explicit-agreement and other training rules described in other Cursor documents is not stated on this page.
Sharing & Disclosure · Company practice
With Privacy Mode off, some inference providers may temporarily store model inputs and outputs, deleted after use
If you turn off Privacy Mode, the page says some of Cursor's inference providers may temporarily access and store model inputs and outputs to improve their inference performance, and that this data is deleted after use.
Supporting quotations
Some of our inference providers may temporarily access and store model inputs and outputs to improve our inference performance; this data is deleted after use.
Not stated here
Which providers, and what “deleted after use” means in terms of time, are not stated.
Use & Processing · Company practice
Requests go through Cursor's backend even if you use your own API key
Even if you use your own API key, your requests still go through Cursor's backend, where it does its final prompt building.
Supporting quotations
Even if you use your API key, your requests will still go through our backend!
where we do our final prompt building.
Not stated here
What Cursor's backend retains of such requests is not stated in this note; the Privacy Mode statements above address training and storage.
Retention & Deletion · Company practice
File contents are cached temporarily on Cursor's servers, encrypted, and never used as training data in privacy mode
Cursor temporarily caches file contents on its servers to reduce latency and network usage. The files are encrypted using unique client-generated keys, and those keys exist on its servers only for the duration of a request. All cached file contents are temporary, never permanently stored, and never used as training data when privacy mode is enabled. The page does not say whether cached contents may be used for training when privacy mode is off.
Supporting quotations
We temporarily cache file contents on our servers to reduce latency and network usage.
The files are encrypted using unique client-generated keys, and these encryption keys only exist on our servers for the duration of a request.
All cached file contents are temporary, never permanently stored, and never used as training data when privacy mode is enabled.
Qualifications & exceptions
The no-training statement is made for when privacy mode is enabled.
Not stated here
How long “temporary” is for cached file contents is not stated.
Whether cached contents may be used for training when privacy mode is off is not stated.