ChatGPT Enterprise and alternatives

Is ChatGPT Enterprise Safe for Confidential Client Work, and What Are the Alternatives?

Updated November 13, 2026

Short answer: ChatGPT Enterprise does not train on your data by default and lets your admins control retention, which is a real step up from consumer ChatGPT. But it is not zero data retention: your chats are stored until they are deleted, and deleted chats can take up to 30 days to leave OpenAI's systems. The same is true, in different ways, of Claude Enterprise and Microsoft 365 Copilot. Whether that is safe enough depends on your matters. If you want a tool that does not store client content, the alternatives are a tool that holds content only in memory until it is reset, such as Faradex, a privacy-first chat tool such as Proton Lumo for Business, or zero data retention through an API, which takes technical work.

"Doesn't ChatGPT Enterprise already give me zero retention?"

This is one of the questions we hear most from firms. The short answer is no. Zero data retention, where OpenAI offers it, is a setting for its API, the programming interface developers use, not for the ChatGPT apps your lawyers log in to. OpenAI's API offers it on eligible endpoints "subject to prior approval by OpenAI". The chat products, including Enterprise, keep your conversations.

What ChatGPT Enterprise does offer is control and a no-training default. That is valuable. It is just a different promise.

ChatGPT Enterprise and ChatGPT Business, in OpenAI's words

These terms come from OpenAI's own pages, read on November 11, 2026 (pricing, enterprise privacy).

ChatGPT Business ChatGPT Enterprise
Who it is for "Business plans are available starting at 2 users." "Please contact sales to learn more about our Enterprise plan."
How you buy it Self-serve, with monthly or annual plans Through OpenAI's sales team, on annual plans
Training "By default, we do not use your business data for training our models." The same default
How long chats are kept "Any deleted or unsaved conversations are removed from our systems within 30 days, unless longer retention is required by law, or is reasonably necessary to protect our services or any third party from harm." "Your workspace admins control how long your data is retained. Any deleted conversations are removed from our systems within 30 days, unless we are legally required to retain them."
Automated review "We may run any business data submitted to OpenAI's services through automated content classifiers and safety tools" The same statement covers business data
Zero data retention Not for the ChatGPT apps Not for the ChatGPT apps

Read the two retention lines closely. Enterprise's exception is narrow: "unless we are legally required to retain them." Business adds a second one: "or is reasonably necessary to protect our services or any third party from harm." OpenAI also says "ChatGPT Business successfully completed a SOC 2 Type 2 audit"; that is OpenAI's own statement.

ChatGPT Enterprise and Claude Enterprise both say they do not train on your business data unless you opt in.

What the NYT v. OpenAI order showed about stored chats

In the copyright case brought by The New York Times, a court ordered OpenAI in May 2025 to preserve user data it would normally delete. OpenAI says "Our obligations under the earlier order ended on September 26, 2025", and that it still holds limited historical user data from April to September 2025 (OpenAI). The broad order ended then, though data already kept, and some named accounts, stayed under preservation (Duane Morris).

Who was covered, in OpenAI's words: users of "ChatGPT Free, Plus, Pro, and Team subscription or if you use the OpenAI API (without a Zero Data Retention agreement)." Team was OpenAI's business plan at the time. Not covered: "ChatGPT Enterprise or ChatGPT Edu customers", and API customers with zero data retention, about whom OpenAI said "we never retain the prompts you send or the answers we return. Because it is not stored, this court order doesn't affect that data."

The lesson is not that Enterprise is unsafe; Enterprise was outside that order. It is that a court can order a provider to keep what it would normally delete. OpenAI says the order did not affect zero data retention customers, because that data was not stored. A preservation order applies to data a provider keeps; zero data retention means there is less kept data for an order to reach. The subpoena page takes this further.

Claude Team and Claude Enterprise

Anthropic's chat products follow the same pattern (Claude pricing, Enterprise retention, zero data retention).

  • Price: Anthropic's FAQ says Claude Enterprise "Pricing is $20 per seat per month plus usage billed at API rates", that "Enterprise is billed annually", and that "You can get started self-serve or contact sales for a tailored quote."
  • Retention on Enterprise: "By default, data is retained indefinitely unless a custom retention period is set." "The minimum retention period is 30 days", and "by default, projects are retained indefinitely". Custom retention is available to Enterprise customers.
  • Zero data retention: "Claude Teams and Claude Enterprise product interfaces: These interfaces are not ZDR-eligible." The one exception Anthropic names is Claude Code on Claude Enterprise, "available only to qualified accounts". For its APIs, "To request ZDR for your organization, contact the Anthropic sales team. ZDR is enabled per organization."
  • Training: Anthropic says commercial data is "never used for model training without your express permission."

So for a firm that wants a chat product, Claude's zero data retention is API-only and goes through sales.

Microsoft 365 Copilot

For a firm already on Microsoft 365, Copilot is the closest option to hand (Microsoft, Microsoft Learn).

  • Price: the Copilot Business add-on is shown as "Originally starting from $21.00 now starting from $18.00 user/month, paid yearly", a promotion "available between July 1, 2026, and December 31, 2026" and "available only to existing Microsoft 365 customers with an eligible Business plan". "An annual commitment is required." Bundles: Business Standard with Copilot at $23.50 and Business Premium with Copilot at $32.00 per user per month, paid yearly. Business plans cover "up to 300 users".
  • Training: "Prompts, responses, and data accessed through Microsoft Graph aren't used to train foundation LLMs".
  • Retention: "The stored data includes the user's prompt and Copilot's response ... This data is processed and stored in alignment with contractual commitments with your organization's other content in Microsoft 365. The data is encrypted while it's stored". Admins can use Microsoft Purview to set retention, and users can delete their activity history.
  • Where: "Customers outside the EU may have their queries processed in the US, EU, or other regions."
  • A real strength: Copilot works inside your existing Microsoft 365 permissions; it "presents only data that each individual can access".

Our reading: because Copilot stores interactions with your other Microsoft 365 content, they are searchable by your admins and within reach of any e-discovery your firm runs or is ordered to run. That can help with record-keeping. It also means AI history is part of what can be produced.

The alternatives, by what they keep

If the stored-chat model does not fit your matters, there are three directions.

A Workspace that holds content only in memory: Faradex

Faradex is a private AI workspace for small and mid-sized firms that handle confidential client work. Each firm gets its own Workspace on dedicated infrastructure in the United States. Files and conversations are held in memory and are gone when the Workspace is reset. AI model: Currently Opus via AWS Bedrock, more information is available on our website. AWS does not store prompts or responses, and model providers have no access to them. Pricing: from $120 a month, no seat minimum; see faradex.ai/pricing.

Privacy-first chat: Proton Lumo for Business

Lumo for Business costs "$11.99 /user per month" billed yearly, with a 14-day free trial. Proton says "After Lumo processes your query and generates a response, the data is erased", and "Lumo doesn't use your chats to train the AI models" (Proton). The catch: it is "powered exclusively by open-source models", not frontier models, and Proton is "Built and based in Europe".

CoCounsel, Clio's built-in AI, Vincent by Clio and Spellbook each have zero retention agreements with the model providers behind them, but store your firm's content themselves. That suits a firm that wants the AI to remember its work, inside a legal product. The main comparison quotes each one's terms.

Zero data retention through an API

OpenAI and Anthropic both offer zero data retention on their APIs, with approval or through sales. That route needs someone to build or run the tool your lawyers use on top of it, so it suits firms with technical staff or a developer they trust.

How to decide

If your main concern is... Consider
A familiar tool with a no-training default and admin controls ChatGPT Enterprise or Business, Claude Enterprise, or Copilot, knowing chats are stored
Staying inside software you already use Copilot in Microsoft 365, or Clio's built-in AI
Legal research and drafting tools that remember your work A legal platform, knowing it stores content and has zero retention only at the model provider
Client content that is not stored A tool that clears client content, such as Faradex (held in memory, gone on reset) or Proton Lumo for Business (erased after each response)

Whichever you choose, the duty to understand the tool's terms is yours. Our companion site, AI Ethics for Lawyers, covers what ABA Formal Opinion 512 asks, and the vendor questions turn it into a list you can send. Comparing legal platforms such as Harvey, CoCounsel and Lexis+ with Protégé? See Harvey AI alternatives for small law firms.

This guide is general information, not legal advice. Check each vendor's current terms on its own pages before you buy.

About Faradex. Faradex is a private AI workspace for small and mid-sized firms that handle confidential client work. Each firm gets its own Workspace on dedicated infrastructure in the United States. Files and conversations are held in memory and are gone when the Workspace is reset. AI model: Currently Opus via AWS Bedrock, more information is available on our website. AWS does not store prompts or responses, and model providers have no access to them. Pricing: from $120 a month, no seat minimum; see faradex.ai/pricing.