OpenAI now lets ChatGPT users turn off their chat histories, excluding them from the company's models, and plans to launch ChatGPT Business in the coming months
here's how to do it Pranav Dixit / Business Today : OpenAI's ChatGPT announces new plan, introduces incognito mode to address privacy concerns Riley Kaminer / Refresh Miami : Lucyd is bringing ChatGPT right into your ears Marcus Gopolang Moloko / Memeburn : You'll soon pay OpenAI, plans for ChatGPT subscription tier brew Emre Çitak / gHacks Technology News : ChatGPT takes a step to ease privacy concerns Damir Mujezinovic / MakeUseOf : Does ChatGPT Have Any Security Issues? Matt Milano / WebProNews : OpenAI Adds Privacy Controls to ChatGPT Ryan Morrison / Tech Monitor : OpenAI to launch ChatGPT for enterprise Fox Business : OpenAI says ChatGPT feature letting users disable chat history now available Cybersecurity Dive : OpenAI adds more data privacy guardrails for ChatGPT Analytics Insight : OpenAI Plans to Launch a New Subscription Tier for ChatGPT Jai Pratap / CoinGape : OpenAI Unveils Latest Features For ChatGPT Cecily Mauran / Mashable : ChatGPT rolls out important privacy options Nikki Main / Gizmodo : How to Hide Your Chats from ChatGPT's Algorithm The Information : OpenAI Teases Business Version of ChatGPT With Privacy Controls Paul Hill / Neowin : ChatGPT lets you go incognito which disables history and the use of your data for training Tweets: @openai : ChatGPT users can now turn off chat history, allowing you to choose which conversations can be used to train our models: https://openai.com/... Sam Altman / @sama : you can now disable chat history (and training) in ChatGPT, and we will offer ChatGPT Business in the coming months. https://openai.com/... Greg Brockman / @gdb : Can now turn off chat history in ChatGPT (which also opts the conversation out of training) and export your data: https://twitter.com/... https://twitter.com/... @dharmesh : I am constantly impressed by how fast @sama and the @OpenAI team move. They keep knocking down the dominos of resistance, one-by-one. https://twitter.com/... Anton / @abacaj : For everyone saying that they need their own models to own their data. If OAI can just guarantee that they won't use your data then it will just be that much harder to justify your own model https://twitter.com/... Ethan Mollick / @emollick : One of the big barriers to adopting ChatGPT for serious use was not knowing whether it was using your data (or your company's data) for training Now, it is much easier to use ChatGPT without worrying about that quite as much (though there are still unknowns in the announcement) https://twitter.com/... Juan Ortiz Freuler / @juanof9 : after scraping the open web and training the chatbot on people's content, without any user's explicit consent: “We want to move more in this direction where people who are using our products can decide how their data is being used” 🫠 https://www.bloomberg.com/... @business : OpenAI is aiming to make people feel comfortable using the chatbot for a wide range of applications. https://www.bloomberg.com/... Rachel Metz / @rachelmetz : it's like incognito mode for chatgpt https://www.bloomberg.com/...
Context & Ripple Effects
OpenAI had already begun commercializing its models through a business-focused ChatGPT API with dedicated capacity, making data handling a practical adoption issue rather than only a consumer concern. The new controls put a user-facing boundary around conversations while the company prepares a separate business offering.
The move also sits at the start of a product arc that later included ChatGPT Enterprise with stronger privacy and data-analysis features. Subsequent removal of a feature that exposed conversations to search engines underscores how disclosure settings can become a recurring trust and product-governance issue.
First-order effects
- ChatGPT users can disable history, keeping new conversations out of model training, and can export their data; OpenAI gives privacy-sensitive users a way to continue using the service with a clearer data-use choice.
- A planned ChatGPT Business product creates a distinct route for organizational customers whose adoption depends on more formal privacy and data-handling assurances.
Second-order effects
- OpenAI must balance the trust benefit of opt-outs against losing some conversational data that could otherwise inform model improvement, increasing the importance of consented and business-supplied usage patterns.
- Competitors selling generative-AI assistants face pressure to make training, retention, and export controls legible to both individual and workplace users, not merely state them in policies.
Third-order effects
- If these controls become standard, conversational AI will increasingly compete on configurable data governance—who can retain, train on, export, or expose prompts—alongside model quality.
- The later reversal of searchable shared conversations suggests that privacy settings in AI products may remain subject to rapid adjustment as providers test distribution features against user expectations and scrutiny.
The trend: Consumer AI is evolving from a default data-collection model toward segmented products in which access and privacy controls are core parts of the offering.