LinkedIn’s use of member data for AI training has put a familiar platform question back in front of users: who gets to decide when posts, activity and feedback become training material for new tools?
The issue centers on a LinkedIn setting available to users in the U.S., but not to users in the EU, EEA, or Switzerland. That setting lets members turn off use of their data for training “content creation AI models.” The toggle existed before LinkedIn updated its terms of service and privacy language to reflect that use.
What LinkedIn changed
LinkedIn has now updated its terms of service. The concern is timing. Normally, a platform changes its terms before using personal data for a new purpose, giving people a chance to review the change, adjust their account settings, or leave the service.
In this case, LinkedIn appears to have made the opt-out available while its policy language had not yet caught up. The situation was first reported by 404 Media, according to the source article.
The LinkedIn setting is found in the “Data Privacy” section of the desktop settings menu. The relevant page is called “Data for Generative AI improvement,” and the option reads “Use my data for training content creation AI models.”
Turning that setting off is not a full reset. LinkedIn notes that an opt-out will not affect training that has already taken place.
What data may be involved
LinkedIn says the AI training can support its own models, including models used for writing suggestions and post recommendations. The company also says generative AI models on its platform may be trained by “another provider,” including its corporate parent Microsoft.
LinkedIn’s Q&A describes a broad set of activity that may be collected or processed when members use the platform. The examples include use of generative AI or other AI features, posts and articles, how often someone uses LinkedIn, language preference and feedback provided to LinkedIn teams.
The company previously told TechCrunch that it uses “privacy enhancing techniques, including redacting and removing information, to limit the personal information contained in datasets used for generative AI training.”
For members, the important practical point is that LinkedIn is not describing this as only a narrow feature interaction. The source material says the company’s explanation includes platform use, content and feedback, all tied to improving or developing LinkedIn services.
Who can opt out
The opt-out setting described in the source applies to LinkedIn users in the U.S. It is not available to users in the EU, EEA, or Switzerland, likely because those regions have different data privacy rules.
Ireland’s Data Protection Commission, described as the supervisory authority responsible for monitoring compliance with the GDPR, told TechCrunch that LinkedIn informed it last week that clarifications to its global privacy policy would be issued today.
The DPC also said LinkedIn advised that the policy would include an opt-out setting for members who did not want their data used to train content generating AI models. According to the DPC spokesperson, the opt-out is not available to EU/EEA members because LinkedIn is not currently using EU/EEA member data to train or fine-tune these models.
LinkedIn also provides a form that users can try for a broader opt-out. But again, the company says opting out does not undo model training that has already happened.
Why privacy groups object
The nonprofit Open Rights Group has asked the Information Commissioner’s Office, the U.K.’s independent regulator for data protection rights, to investigate LinkedIn and other social networks that train on user data by default.
Mariano delli Santi, ORG’s legal and policy officer, criticized the opt-out model directly. In the statement cited by TechCrunch, delli Santi said: “LinkedIn is the latest social media company found to be processing our data without asking for consent,” and argued that “Opt-in consent isn’t only legally mandated, but a common-sense requirement.”
The dispute is not only about LinkedIn. The source article places the company inside a wider shift across online platforms, where user-generated content is increasingly treated as a resource for generative AI development.
Meta announced earlier this week that it was resuming plans to scrape user data for AI training after working with the ICO to simplify the opt-out process. Other networks, including Tumblr owner Automattic, Photobucket, Reddit and Stack Overflow, have licensed data to AI model developers.
Stack Overflow’s experience shows why users may react strongly when platform content is reused. After Stack Overflow announced that it would begin licensing content, some users deleted posts in protest, only to see those posts restored and their accounts suspended.
What users should understand now
For LinkedIn users in the U.S., the immediate action is straightforward: review the “Data Privacy” settings and decide whether to leave “Use my data for training content creation AI models” on or off.
The larger issue is less simple. Opt-out systems put the burden on users to notice each policy shift, understand the consequences and change settings before data is used. That becomes harder as more platforms introduce AI features and reuse existing user content in new ways.
LinkedIn’s updated terms may now document the practice, but the timing of the update is what made the issue notable. For users, the lesson is that AI settings are becoming part of routine account privacy maintenance, especially on platforms built around professional identity, posts, articles and engagement data.