What is data residency? What it covers in AI, and who offers it

The Morse teamPublished November 12, 20268 min read

Data residency is a provider's promise that your data is stored in a specific country or region, such as the EU or the United States, usually one the customer picks. In AI services, the promise covers saved data like chat history and uploaded files. It often leaves out where the model processes your prompt. As of November 12, 2026, OpenAI, Anthropic, Google, and Microsoft sell residency controls to organizations. None of their residency settings reaches a personal account. Mistral is the exception, because it hosts everyone's data in the EU by default.

We read the data residency pages of OpenAI, Anthropic, Google, Microsoft, and Mistral on November 12, 2026, along with the section of the US Code that the CLOUD Act added and the text of the GDPR. Every row in the table below links to the page it came from. The word "residency" means something different at each of these companies, so check what a given offer covers before you rely on it.

Which AI providers offer data residency?

Read on November 12, 2026. "At rest" is where saved data is stored. "Inference" is where the model runs on your prompt.

Only Mistral's row reaches a free or personal account.

What does data residency cover in AI?

Mostly data at rest. Data in an AI service is in one of three states: at rest (saved chats, files, and their backups), in transit (moving between your device and the provider), and in processing (the model working on your prompt). Residency offers are written around the first state, and each provider says separately how far they reach into the other two.

• At rest. OpenAI defines ChatGPT data residency as keeping "in-scope customer content stored at rest in a specific geographic region" (OpenAI). That content includes conversations, files, Memory, custom GPTs, and image generation inputs and outputs, with their backups and replicas kept in the same region. Account data, billing details, and logs without content are out of scope.
• In transit. OpenAI's API terms say residency doesn't apply to "any transmission or storage of Customer Content outside of the selected region caused by the location of an End User" (OpenAI). For its US and EU API addresses, OpenAI uses Cloudflare Regional Services so that "TLS termination and HTTPS decryption occur within the selected processing region."
• In processing. This is where the model runs, and most offers treat it as a separate setting.

Some data leaves the region by design. OpenAI lists web search, apps and MCP connections, workspace metadata, the workspace name, billing information, and user logins as data that may be stored outside the chosen region (OpenAI). Its API documentation adds that residency "does not apply to system data," such as analytics, usage statistics, and support requests (OpenAI). Google's data regions don't cover "logs or cached content" or Gemini Notebook, formerly NotebookLM (Google). When a US user and an EU user message each other directly, Google stores the messages in both regions (Google). Microsoft's EU Data Boundary page says customer data will still leave the boundary in "limited circumstances," which Microsoft documents on separate pages (Microsoft Learn).

Does data residency cover where the AI model runs?

Usually not. OpenAI, Anthropic, Google, and Microsoft each treat storage and processing as separate settings. Most of their residency offers stop at storage. Mistral's subprocessor list gives a location for both. It places its inference provider in the EEA and its own Mistral Compute infrastructure in France, while image generation and web search go to companies in the United States (Mistral Trust Center).

• OpenAI sells in-region GPU work only to Enterprise and Edu, and only in Europe, the US, and the UAE; even then "non-GPU processing may still occur globally" (OpenAI).
• Anthropic works the other way around. Its Enterprise setting "doesn't control data storage," only where inference runs (Anthropic).
• Google Cloud keeps stored data in the location you picked whatever endpoint you call, but a request sent to a global endpoint "may be processed in any Google Cloud location around the world" (Google Cloud).
• Microsoft sends Copilot's model calls to the nearest data centers in the region, but they "can call into other regions where capacity is available during high utilization periods" (Microsoft Learn). Customers outside the EU "may have their queries processed in the US, EU, or other regions."
• Claude inside Copilot is "currently excluded from the EU Data Boundary," and Microsoft turns Claude models off by default for customers in the EU, EFTA, and the UK (Microsoft Learn).

If you're buying for an organization, get both answers in writing: where content is stored at rest, and where inference runs.

Does ChatGPT Plus have data residency?

No. OpenAI offers stored-content residency to eligible API customers and new ChatGPT Enterprise and Edu customers. Its help center names no consumer plan (OpenAI). A February 7, 2025 ITdaily article said "ChatGPT Plus and Edu users can choose to store their data within the borders of Europe" (ITdaily), but OpenAI's own European announcement named only ChatGPT Enterprise, ChatGPT Edu, and the API Platform (OpenAI). For what a personal account gets at each company, see where ChatGPT and 11 other AI apps store your chats.

Mistral is the one provider here that hosts every account in the EU by default, according to its help center dated August 12, 2026 (Mistral), and is Le Chat private covers what it keeps there.

What is the difference between data residency and data sovereignty?

Data residency is where data sits. Data sovereignty is whose laws can reach it. Data localization, a third term, is a law that requires certain data to stay inside a country.

The three come apart whenever a company based in one country stores data in another. A residency setting can meet a localization rule and still leave the sovereignty question open, because the company holding the data may answer to another country's courts.

OpenAI pitched European residency as a way to help organizations "meet local data sovereignty requirements" (OpenAI). US law doesn't stop at the server's border, though. The CLOUD Act added a section to the US Code that requires a provider to preserve or disclose data in its "possession, custody, or control, regardless of whether such communication, record, or other information is located within or outside of the United States" (18 U.S.C. 2713). The Department of Justice says the US enacted the CLOUD Act in March 2018, and that it lets foreign partners with bilateral agreements obtain electronic evidence from US-based providers "wherever it happens to be located" (DOJ). Its page lists agreements with the United Kingdom, signed October 3, 2019, and Australia, dated December 15, 2021.

The EU's text pulls the other way. Article 48 of the GDPR says a court judgment or authority decision "of a third country" requiring a controller or processor to transfer or disclose personal data "may only be recognised or enforceable in any manner if based on an international agreement, such as a mutual legal assistance treaty" (EUR-Lex). How those two rules meet for a given company is a question for a lawyer. This isn't legal advice. Our guide to whether AI chats can be subpoenaed covers how courts have reached chat logs so far.

How Morse keeps your chat history on your device

Every residency offer above is a promise about which of a provider's servers holds your history. For your history, Morse's answer is simpler: it isn't on our servers at all. Chats, Studio images, characters you make, and settings are kept in your browser, encrypted. Our servers don't keep messages or log what you write, and the browser copy takes two keys to open: one that never leaves your device, and one our server hands you at sign-in.

What leaves the device is the message itself, on its way to a model. If a law or contract requires a region for processing, check with us at hello@trymorse.ai before relying on any mode. Otherwise, pick the mode by who you'd rather not see or keep the message:

• Private, when the worry is a stored copy on someone else's disk. New chats start here. Morse has zero data retention agreements with its inference, search, and other providers, so open-weight models like DeepSeek, Kimi, and GLM, plus xAI's Grok, answer without anyone keeping the prompt. Where a provider confirms zero retention on each reply, Morse checks it and refuses a reply that doesn't.
• E2EE, when the worry is who can read the prompt while it's processed. Your browser verifies the enclave's hardware attestation and encrypts the message for that enclave alone, so Morse's servers relay ciphertext they can't read. The enclaves are run by third-party providers, not by Morse, and the protection rests on the enclave hardware and attestation working as designed. It comes with Pro, at $19 a month.
• Anonymous, when the job needs GPT, Claude, or Gemini. Our proxy strips your account, email, and IP address and sends the request under Morse's own account. The provider sees the message without knowing it's you, and may keep it for a limited time to check for abuse, under its own policy.

Morse doesn't train AI on your content, sell your data, or show ads (Morse privacy policy), and because we don't store chats, we usually have very little to give authorities. Your history is kept in your browser, so clearing the browser or deleting your account removes it for good, and Morse can't recover it. Keep a copy of anything you need.

You can send a first message from the Morse home page without an account, or start free with no card needed.