Most Major AI Chatbots Offer Training Opt-Outs; US Meta Users Do Not

Opting out does not apply retrospectively, so old chats can still be used, and it does not delete any data, which remains subject to each platform's retention policies. Personalization and memory are managed by separate settings, so turning off training use should not affect day-to-day use, Engadget said.

For ChatGPT, OpenAI's opt-out toggle is called Improve the model for everyone, found under Settings and then Data Controls, and all personal user accounts are opted in by default. Opting out applies to the entire account across web and mobile, but conversations still appear in chat history. ChatGPT also offers Temporary Chats, which are never used for training and are purged by OpenAI after 30 days.

Anthropic did not use consumer chats for model training until August last year, when it rewrote Claude's terms and gave every Free, Pro and Max user a deadline of September 28 to accept or decline. Claude's toggle is under Settings and then Privacy. If it remains on, Anthropic keeps conversations for up to five years; switching it off returns accounts to the older 30-day deletion window. Opting out stops future use, but Anthropic keeps whatever it already trained on.

Google's control was recently renamed Keep Activity, and it is also on by default. When enabled, conversations can be used to improve Google's models, and a subset of chats are read by human reviewers. Any chat a reviewer accesses is retained for up to three years, even after the user deletes history. Switching off the setting ends training use and stops interactions from being saved to Gemini's chat history, but Google still retains each conversation for 72 hours to run the service, using the last 24 hours for contextual responses.

Microsoft has two controls for model training on user data: Training on conversation activity, which is on by default, and Training on voice conversations, which is off by default. Both are under the profile icon, then Settings and then Privacy. Opting out only excludes future conversations from model training and does not stop Microsoft from using chats for advertising or general product improvements.

X enrols every account in Grok training by default, covering both public posts and conversations with SpaceXAI's chatbot. The toggle is under Settings and privacy, then Privacy and safety, then Grok & Third-Party Collaborators. Unchecking the data-sharing option withdraws posts and Grok interactions from future training and fine-tuning.

Meta does not allow US users to opt out at all. People in the UK and EU are protected by strong data privacy laws under the EU General Data Protection Regulation, and they can object to their information being used for model training through a form in Meta's privacy center. Since last December, text and voice chats with Meta AI have been used to feed the company's advertising machine across Facebook, Instagram, WhatsApp and Messenger. For US users, the only option is to not engage with Meta AI in the first place, which the report says is not terrible advice following the launch of Meta's new AI agent, Muse.