Editorial status: 12 September 2026. This overview describes documentation, not practical account testing. Sources state the date and scope of the review. Before use, check the current state of your own account, region and mode.
ChatGPT: memory and Temporary Chat
Depending on settings, a new chat may use information from the past. Memory is managed separately from history and from using data to improve models. Switching off memory does not itself delete already stored information; when removing a detail, also check its source chats and files.[1][2]
The newer FAQ and announcement of 27 August 2026 distinguish Temporary Chat without personalisation from a version using existing memory, instructions and connections. Temporary Chat does not create new memory; saving it converts it to an ordinary chat with the relevant account rules. While it remains temporary, it is not used to improve models and is not in history, but a copy may be kept for up to 30 days for safety purposes.[3]
When reviewing a conversation, choose an available mode without personalisation. Some help pages loaded still contained an older description, so check your account’s actual options. Temporary does not mean anonymous, and data passed to connected services follow their own terms.[3:1]
Training data and feedback
In ChatGPT’s data controls, Improve the model for everyone can be switched off. According to the documentation, new conversations are then not used to improve models. This does not end history or operational and safety processing.[2:1]
For consumer Claude, use for model improvement depends, among other things, on the user’s choice. Safety checks and providing feedback have special rules. According to the documentation, submitting a thumbs-up or thumbs-down may include the entire related conversation, retention for up to five years and use for training. The same source says Incognito mode is not used to improve Claude. A paid consumer account is not the same as a business or developer offering.[4]
Parental settings and safety notifications
ChatGPT’s parental features allow selected functions and quiet hours to be managed, not the child’s conversations to be read. The documentation describes a notification when a teenager disconnects the account, and model-improvement use being off by default for conversations in a linked teen account. Some changes take effect only in new chats. These options do not provide complete oversight of a child’s safety.[5]
Trusted Contact is an optional feature for adult personal accounts in supported countries. The chosen person must accept the invitation. For a defined serious suicide-related concern, after automated assessment and human review, they may receive a brief notification, not the full transcript. Check availability for your account and communication mode in current help. The feature can miss a situation, and the contact may be unavailable. If you are concerned about safety, seek help directly.[6]
Sources and notes for this chapter
OpenAI. Memory FAQ: help.openai.com/en/articles/8590148-memory-faq. State reviewed 10 September 2026. Documentation of memory and its management, not a study of clinical understanding of a user. ↩︎
OpenAI. Data Controls FAQ: help.openai.com/en/articles/7730893-data-controls-faq. Documentation state reviewed 11 September 2026. Concerns data-use settings in ChatGPT; these are not identical to the rules of every OpenAI service. ↩︎ ↩︎
OpenAI. Temporary Chat FAQ and ChatGPT Release Notes, entry of 27 August 2026. FAQ, release notes. A newer indexed version with personalisation and saving options was available during review on 12 September 2026. Other help pages contained an older description. Using existing memory is not creating new memory; availability of the mode in a particular account was not tested. ↩︎ ↩︎
Anthropic. Is my data used for model training?: privacy.claude.com/en/articles/10023580-is-my-data-used-for-model-training. State reviewed 11 September 2026. Consumer accounts, consent, safety exceptions and feedback. When the offering changes, check the terms of the particular account. ↩︎
OpenAI. Parental controls in ChatGPT: help.openai.com/en/articles/12315553-parental-controls-in-chatgpt. State reviewed 11 September 2026. Scope of account control, conversation privacy and limitations of notifications. ↩︎
OpenAI. Trusted contacts in ChatGPT. Current overview. Available official wording reviewed 12 September 2026: adult personal accounts in supported countries, accepting an invitation, a defined notification after human review and service limitations. The overview did not confirm a previously stated detail about Voice; the text therefore does not adopt it as a current rule. Availability in accounts was not practically tested. ↩︎