Skip to Content
Platform.Ontology

Data for Generative AI Improvement setting

A LinkedIn account setting that controls using member content to train AI models.

Works without askingContacts and personal dataProfile and content dataAndroid and iPhoneWhen you are signed in

In one line

Can use your posts and profile to train AI unless you switch the setting off

A LinkedIn account setting that controls using member content to train AI models.

Posts, profile data and job application data can train content-generating models. The toggle is forward looking; it cannot undo past training.

What it can and can't do

What the user does: You open Settings and turn Data for Generative AI Improvement on or off

It can

  • Use posts, profile and job application data to train content making AI
  • Follow your setting from the time you switch it, in the app and on the web

It cannot

  • Undo training that already happened when you switch it off
  • Use your data for training after you turn it off, going forward

On Android and iPhone

AndroidiPhone
StatusSupportedSupported
First versionNot availableNot available
Permission or entitlementNot listedNot listed
Protection levelnonenone
How access is grantedsettingssettings
What the user seesSetting in the app and on the webSetting in the app and on the web
How long it lastsUntil changedUntil changed
How to take it backProfile > Settings > Data privacy > Data for Generative AI ImprovementProfile > Settings > Data privacy > Data for Generative AI Improvement
Store policyNone listedNone listed
APIAccount settingAccount setting
NotesNoneNone

Data you get

What comes back: Use my data for training content creation AI models
Shape: boolean
How often: per account

No individual data fields. The signal is the access itself.

Technical limits

  • Does not cover feedback data; a separate objection form exists for other AI uses

The story

What it is

A LinkedIn setting that decides whether your posts and profile can train AI models.

For you

In your LinkedIn privacy settings there is a switch for using your data to improve generative AI. It is on by default in many regions. You can turn it off, but past use is not undone.

For makers

AI features need lots of real text. Member posts, profiles and job info are rich training material. The setting lets LinkedIn say people can opt out, while it keeps the data flowing by default.

Where you see it

TypeExampleApps
productivityA writing assistant drafts a LinkedIn post using models trained on member content.LinkedIn
enterpriseA recruiter tool suggests job descriptions learned from real postings.LinkedIn
educationA resume helper learns what good profiles look like.None named
adsAn ad tool writes copy tuned to what members respond to.None named
securityA fraud team trains spam and fake profile detectors.None named
analyticsA skills model learns which jobs link to which skills.None named
socialA feed model learns what content people like.None named

Risks

  • Past training not undone
  • On by default
  • Profile text may appear in outputs

Android and iPhone in short

  • Android: The setting works the same in the Android LinkedIn app.
  • iPhone: The setting works the same in the iPhone LinkedIn app.

Good uses and abuse

Legitimate uses

  • Writing assistants

Known abuse

  • Training on private content

Scores

ScoreValueMeaning
Sensitivity4 of 5How revealing the data is
Utility3 of 5How useful it is to honest apps
Friction3 of 5How much the user has to do
Fingerprint bitsNot ratedRough identifying entropy, when known

A real example: LinkedIn

Default on for members outside excluded regions from Sept 2024 (opt-out was hidden). From 3 Nov 2025 the updated notice applies to Designated Countries and UK; LinkedIn says opting out is forward looking. Same setting in mobile Settings; the help page has no mobile specific steps.

History

  • 2024-09, Android n/a: Setting appeared on by default for most regions; EEA and Switzerland excluded; UK paused after ICO contact
  • 2025-11, iPhone n/a: Setting extended to EU, EEA, UK, Switzerland, Canada and Hong Kong from 3 Nov 2025 under the updated privacy notice

Sources

How well checked: Checked against real behavior.

On this page