September 28, 2026
Data can be excluded from AI training: 6 services offer a separate switch
As of Sep 29, 2026, six services let users opt out of training on future data. In ChatGPT, this choice preserves chat history and applies to new Codex tasks on the personal plan.
In ChatGPT, open Profile → Settings → Data Controls and turn off “Improve the model for everyone.” Temporary Chat will not appear in history or training, and OpenAI will delete it after 30 days. In Claude, go to Settings → Privacy → “Help Improve our AI models” → Off. Claude will not add new chats and coding sessions to future training runs.
Other services. In Gemini, open Settings and help → Activity and choose Turn off or Turn off and delete activity. Google keeps chats for up to 72 hours after the setting is disabled and deletes activity after 18 months by default. In Microsoft Copilot, turn off Training on conversation activity and Training on voice conversations in Privacy. In Grok, revoke Data Sharing permission in Grok & Third-party Collaborators. In LinkedIn, turn off Data for Generative AI Improvement, which is enabled by default.
Before the setting is disabled, new data is handled under each service’s rules. After it is disabled, ChatGPT, Claude, Copilot, Grok, and LinkedIn exclude future conversations or content from training. This choice does not undo Claude training already in progress or models that have already been trained.
For Claude, conversations flagged by the safety classifier may still be used in internal trust and safety models.