Tech Giants Added AI Data Opt-Out Controls
Major platforms introduced settings to restrict user data from training, as EU regulatory debates on data rights intensify.
Updated on Oct. 9, 2026 in Artificial Intelligence
Live Poll
Should companies be required to get your permission before using your personal data for AI training?
OpenAI, Google, Meta, and LinkedIn have implemented user controls allowing individuals to opt out of having their personal data used to train artificial intelligence models. These settings vary by platform and do not retroactively remove data from models already trained on that information.
Why it matters
Companies maintain that utilizing user data for model development constitutes a legitimate interest under GDPR, prompting ongoing legislative negotiations in the European Union. These discussions aim to establish whether users should have an unconditional right to object to their data being used in this manner.
Google retains user activity data for 72 hours after settings are disabled, while OpenAI stores temporary chats for 30 days and retains data seen by human reviewers for up to 3 years. Completing the opt-out processes across all four major platforms requires approximately 15 minutes.
The players
OpenAI
Developer of the GPT series of large language models that uses consumer interactions to refine model performance.
Alphabet subsidiary that operates the Gemini AI models and maintains extensive user activity logging across its ecosystem.
Meta
Parent company of Facebook and Instagram that leverages public user-generated content to train its Llama suite of AI models.
Professional networking platform owned by Microsoft that recently began incorporating member data into its generative AI training pipeline.
The details
Platforms utilize different mechanisms to control data usage, ranging from dedicated settings toggles to privacy objection forms. OpenAI relies on a specific setting to 'Improve the model for everyone,' while Meta forces users to submit forms through a unified Accounts Center that links to Facebook and Instagram. Google tracks activity by default for users 18 and older, requiring manual intervention to disable. These systems operate on the principle of 'legitimate interest' under the General Data Protection Regulation (GDPR), a set of European privacy rules governing how companies process and protect personal data.
Timeline
November 3, 2025: LinkedIn began using European member data for generative AI training.
November 2025: The European Commission proposed an unconditional right to object.
September 3, 2026: The Irish Council presidency drafted an internal compromise text.
Early 2027: A Parliament committee plans to vote on the Digital Omnibus reform package.
The Tech Race
This move follows the European Union's ongoing struggle to define the boundaries of the General Data Protection Regulation (GDPR) in the age of generative models. It represents a shift from passive data harvesting toward a reactive compliance model as the Digital Omnibus reform approaches its first parliamentary hurdle.
Users can now restrict their data from future training, but these settings are not retroactive and do not remove information already incorporated into existing models. Individuals must manually adjust these toggles or submit forms across every distinct platform, a process taking about 15 minutes.
The takeaway
The move to allow data opt-outs reflects a reactive shift by major companies before formal European legislation potentially forces their hand. Watch for the 2027 Digital Omnibus vote, which will clarify whether 'legitimate interest' remains a valid legal defense for AI training.
What happens next
A European Parliament committee is scheduled to vote on the Digital Omnibus reform package in early 2027, which could legally mandate an unconditional right to object to AI training.
Further reading
For a broader look at how these frameworks affect the industry, visit the Artificial Intelligence section.
Source note: This article includes information reported by Notebookcheck.
Live Poll
Should companies be required to get your permission before using your personal data for AI training?
