Anthropic now offers a way for users to transfer saved ChatGPT memory into Claude with a copy-and-paste action, a process the company describes as taking about 60 seconds. The feature, called Import Memory, directly targets people who have built up personalized context inside OpenAI’s chatbot but want to try a competitor without starting from scratch. It’s a vendor-supported option designed to reduce the friction of switching assistants, and it signals a new front in the battle for user loyalty among AI companies.
What ChatGPT Memory Actually Stores
Before transferring anything, users need to understand what they are moving. ChatGPT’s memory system stores saved details and references to chat history that accumulate over time. These can include a user’s name, job title, preferred coding language, dietary restrictions, or any other detail the model picks up during conversations. OpenAI gives users controls and toggles to manage what the system retains, meaning the memory file is not a raw dump of every conversation but rather a curated set of facts the chatbot has been told or has inferred.
That distinction matters because what gets exported is not a full transcript of past chats. It is a condensed list of personal details and preferences. Think of it less like moving an entire email archive and more like handing a new colleague a cheat sheet about how you like to work. The practical result is that Claude will not replicate every nuance of a long ChatGPT relationship, but it will absorb the key reference points that shape how the assistant responds. Users who have invested time teaching ChatGPT about recurring projects, accessibility needs, or household routines can therefore avoid retyping those basics when they experiment with another system.
How the Import Memory Feature Works
The transfer process is straightforward by design. According to Anthropic’s support guide, Claude can import memory from other assistants, explicitly naming ChatGPT and Gemini as compatible sources. Users first export their memory data from ChatGPT’s settings, which produces a text-based list of stored facts. They then bring that data into Claude either by pasting it directly into a conversation or by uploading a text file, so the same workflow applies whether the memory export is short or spans dozens of lines.
Anthropic’s documentation provides specific phrasing users can follow when initiating the import, reducing guesswork about how to present the data. Once the text is submitted, Claude processes the entries and updates its own memory accordingly, mapping each imported fact into its internal profile for that user. The company’s Import Memory page frames the entire workflow as requiring just one copy-and-paste step, reinforcing the 60-second promise in its marketing. There is no API integration, no third-party tool, and no technical setup required. The barrier to entry is essentially zero for anyone comfortable navigating settings menus in two browser tabs and moving a text snippet from one window to another.
Editing and Organizing Imported Memories
Raw transfer is only half the story. One of the more practical aspects of Claude’s implementation is that users can see and edit what the assistant remembers after an import, according to the Import Memory product page. This means if ChatGPT stored an outdated job title or an incorrect preference, the user can correct it inside Claude rather than carrying over stale data. That level of transparency is not just a convenience feature; it addresses a real concern about AI memory systems operating as opaque black boxes where people have limited visibility into what shapes their experience and few tools to fix it.
After importing, users can review what Claude saved and edit or remove items so outdated or incorrect details don’t carry over. In practice, it’s worth scanning the imported list and cleaning it up so the assistant’s saved context stays relevant to how you plan to use Claude.
Why Portability Changes the Competitive Equation
The real significance of this feature is not the 60-second workflow itself but what it represents about switching costs in the AI assistant market. Until now, the accumulated context inside a chatbot acted as a soft lock-in mechanism. Users who had spent weeks or months training ChatGPT to understand their preferences faced a cold start if they wanted to evaluate Claude or any other competitor. Every new tool meant re-explaining the same details, which created friction that favored the incumbent and made experimentation feel expensive in terms of time and attention.
By building a direct import pipeline, Anthropic is deliberately lowering that barrier. The strategic logic is clear: if context portability removes the main reason people stay with a competitor out of inertia rather than satisfaction, then the competition shifts entirely to model quality, safety, and feature depth. That is a fight Anthropic apparently wants to have. In the support materials cited here, OpenAI describes how ChatGPT memory works and how users can manage it, but it does not describe a reciprocal “import memory from Claude” workflow. Over time, if more vendors support similar portability options, users may come to expect that their AI “profile” is portable in the same way that contacts or calendar data can move between apps.
Limits Users Should Know Before Switching
For all its simplicity, the import process comes with caveats that are easy to overlook. The transfer moves saved memory entries, not the conversational style, tone calibration, or behavioral patterns that develop through extended use of any AI assistant. A user who has fine-tuned ChatGPT’s voice through hundreds of interactions will find that Claude responds differently even with the same factual context loaded. Memory import gives Claude the “what” but not the “how” of a prior relationship with another AI tool, so people should expect a period of adjustment as they nudge Claude toward their preferred style.
There is also no publicly available data on how reliably the import process handles edge cases, such as memories stored in non-English languages, heavily technical jargon, or contradictory entries. Anthropic’s documentation describes the workflow and provides user-facing instructions, but independent testing of accuracy rates or error frequencies has not been published. Users making the switch should plan to review their imported memories and clean up any entries that did not translate well, treating the import as a strong starting point rather than a perfect mirror. The feature does accomplish what the headline promises: a fast, low-effort way to carry personal context from one AI assistant to another. Whether that context proves sufficient to make Claude feel like a true continuation of a ChatGPT relationship will depend on how much of that relationship was rooted in raw facts versus the subtler patterns that no export file can fully capture.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
