Claude data privacy on Team: training, retention and access (2026)

Last updated: November 6, 2026

TL;DR

On the Claude Team plan, Anthropic doesn't train on your company's chats by default, keeps them until someone deletes them, and erases deleted chats from its back end within 30 days. For most small businesses that settles the Claude data privacy question. Custom retention, audit logs, HIPAA and zero data retention are the parts that need Enterprise.

• 🚫 Training. Anthropic doesn't train on Team inputs or outputs by default. The exception is feedback: a thumbs up or down sends the whole conversation, and Anthropic keeps it for 5 years. An Owner can switch rating off.
• ⏳ Retention. Chats stay in the product until deleted, and Anthropic publishes no fixed period for Team. Deleted chats leave its back end within 30 days. Incognito chats are kept for 30 days, and chats flagged for a Usage Policy violation for up to 2 years.
• 👀 Access. Your company is the data controller. The Primary Owner can export every member's conversations, incognito chats included. Owners can't view a member's memories in the product.
• 🛡️ Certifications. Anthropic lists SOC 2 Type I and Type II, ISO 27001:2022 and ISO/IEC 42001:2023 for its commercial products, which include Claude for Work.
• 🏢 Enterprise and ZDR. Custom retention, audit logs, the Compliance API and HIPAA need Enterprise. Zero data retention is never a Team option, and even on Enterprise it covers Claude Code, not chat or Cowork.

Claude data privacy and model training

Anthropic doesn't train its models on what your team types into Claude Team, or on what Claude writes back. Its privacy center puts it this way: "By default, we will not use your inputs or outputs from our commercial products (e.g. Claude for Work, Anthropic API, Claude Gov, etc.) to train our models." Claude for Work is Anthropic's name for Team and Enterprise.

Personal accounts work differently. On Free, Pro and Max, training depends on a setting each person controls. Anthropic's data usage docs say it "will train new models using data from Free, Pro, and Max accounts when this setting is on." A teammate doing client work on a personal Pro account is on consumer terms, whatever your company's plan, which is one reason to move everyone onto the Claude Team plan.

There are two ways business data can still reach training. The first is feedback. Per the privacy center's training article: "If you explicitly report feedback or bugs to us (e.g. via our thumbs up/down feedback button), or otherwise choose to allow us to use your data, then we may use your chats and coding sessions to train our models." A rating stores "the entire related conversation" for up to 5 years, de-linked from your user and customer IDs. It doesn't include raw content from connectors such as Google Drive unless someone pasted that content into the chat. That matters more once you connect finance and CRM tools, as the Claude for Small Business plugin does.

The second is opting in, through Anthropic's Development Partner Program, which is available only on its first-party API. You can close the feedback route from two places:

How long Anthropic keeps Team data

Anthropic keeps Team chats in the product for as long as you keep them, and doesn't state a fixed retention period for Team. When someone deletes a chat, it is "Removed from your chat history immediately" and "Deleted from our back-end storage systems within 30 days." Feedback, incognito chats, policy flags and Claude Code each follow their own clock.

Every period Anthropic publishes for a Team account, from its articles on organization data, incognito chats and memory and the Claude Code docs:

Two catch-alls sit over the whole table. Anthropic "may retain your chats or sessions as required by law or as necessary to combat violations of our Usage Policy." And "If permitted in your contract with us, we may anonymize your organization's data for research or statistical purposes," in which case it can keep that data longer. Whether yours permits it is in the contract, not the help center.

The copy most owners forget is on the laptop. Claude Code stores session transcripts "locally in plaintext under ~/.claude/projects/ for 30 days by default," and the cleanupPeriodDays setting changes that. Sessions last continued in Claude Desktop or Cowork are exempt from that limit by default. If your developers share one Claude Code setup, as in our compound engineering for Claude Code post, put the setting there.

Who in your company can see what

Your company owns the data, and the Primary Owner controls it. Under Anthropic's Commercial Terms the customer is the "Controller" and Anthropic is the "Processor", per its processor or controller article. In practice the Primary Owner can export every member's conversations, incognito chats included, but no owner can open a member's memories in the product.

Anthropic's article on who owns team data says those exports "may contain your conversations with Claude, uploaded files, and usage patterns," and that the consumer Terms and Privacy Policy "do not apply to the use of the Claude for Work plan where Anthropic acts as the data processor."

Three things members often get wrong:

• 🕶️ Incognito hides a chat from the company. It hides it from chat history and memory. On Team, incognito chats "are included in organizational data exports" and are kept for 30 days for safety.
• 🧠 Memory is invisible to admins. "Owners can't view or edit a user's individual memories," per the memory article, but memory entries are included in conversation history exports.
• 🗑️ A deleted chat can be recovered by the company. It can't. Messages, files and Projects deleted before an export "will not be included in data exports initiated after the deletion."

The owner's controls sit on one console page, Data and privacy, which in our own Team console is between Usage and Capabilities in the sidebar. The settings, per Anthropic's articles on organization exports, rating chats and project sharing:

Turning off Share projects revokes nothing already shared, but public Projects become private.

Certifications Anthropic lists for Claude for Work

Anthropic's help center lists four compliance credentials for its commercial products, which include Claude for Work: SOC 2 Type I and Type II, ISO 27001:2022, ISO/IEC 42001:2023, and a HIPAA-ready configuration with a Business Associate Agreement. The page doesn't break them out by plan, and the HIPAA piece is Enterprise-only.

That list is from the certifications article. Copies of the reports come from Anthropic's Trust Center, which is where to send a client or insurer who asks for your AI vendor's SOC 2 report.

Claude zero data retention and other Enterprise-only controls

Team has no retention setting, no audit logs, no Compliance API and no HIPAA option. Claude's pricing page lists all four as Enterprise features. Zero data retention isn't a Team option either: even on Enterprise, Anthropic approves it per organization through sales, and it covers Claude Code and the API, not chat or Cowork.

The sources are Claude's pricing page, the custom retention article, the audit logs article and the HIPAA-ready Enterprise article. That last one is blunt: "Team plans and individual plans (Free, Pro, and Max) can't enable HIPAA."

Zero data retention is the most misread term here. Anthropic's Claude Code ZDR docs say it "is not included in the standard Claude for Enterprise plan and cannot be enabled from your admin settings." Its ZDR privacy article limits it to "eligible Anthropic APIs, Anthropic products that use your Commercial organization API key (including Claude Code accessed via the API), and Claude Code for Enterprise plans." Four limits apply even when you have it:

1. It doesn't cover chat on claude.ai, Cowork, or data that passes through third-party tools and MCP servers.
2. Anthropic still keeps User Safety classifier results, and can keep data "where needed to comply with law or combat misuse or harm."
3. It's approved one organization at a time, and doesn't carry over to new organizations under the same account.
4. Features that need server-side storage, such as Claude Code cloud sessions, are switched off.

So the split is simple. Team covers "don't train on our data" and "let us delete it." Enterprise adds "delete it on a schedule" and "show us who did what." ZDR is for developers and API workloads that must leave nothing stored at Anthropic.

Frequently asked questions

Does Claude Team plan train on your data?

No, not by default. Anthropic doesn't use inputs or outputs from Team and Enterprise to train its models. The exception is feedback: if a member clicks thumbs up or down or reports a bug, Anthropic may use that conversation for training. An Owner can turn off the Rate chats setting.

Can my company see my Claude chats?

Yes, through an export. On Team, the Primary Owner can export the organization's data, which includes members' conversations and incognito chats. Owners can't view or edit your memories in the product, but memory entries are included in conversation history exports.

Does Claude keep your data forever?

Not after you delete it. On Team, chats stay in your history until someone deletes them, and Anthropic states no fixed limit. Deleted chats leave its back end within 30 days. Feedback is kept for 5 years, and chats flagged for policy violations for up to 2 years.

How do I prevent Claude from retaining my data?

On Team, delete the chat: it leaves your history at once and Anthropic's back end within 30 days. Incognito chats are kept for 30 days for safety. An organization-wide retention period needs Enterprise, and zero data retention needs an approved Enterprise or API agreement.

Are incognito chats private on Claude Team?

Private from your chat history and memory, not from your company. On Team, incognito chats are kept for 30 days for safety and are included in the organization's data exports. They aren't used for training.

Does Claude Team have zero data retention?

No. Zero data retention is available only to approved API customers and to Claude Code on qualifying Enterprise plans, by request to Anthropic's sales team. Even then it doesn't cover chat on claude.ai or Cowork.

Is Claude Team HIPAA compliant?

No. Anthropic's HIPAA-ready offering, with a Business Associate Agreement, is for Enterprise plans only, and its help center says Team plans can't enable HIPAA. Anthropic says protected health information should only go through features covered by a BAA, and Team has no BAA option.

How secure is the Claude Team plan?

Anthropic lists SOC 2 Type I and Type II, ISO 27001:2022 and ISO/IEC 42001:2023 for its commercial products, including Claude for Work. Team adds SSO and admin controls. Audit logs, custom retention and the Compliance API are Enterprise-only.

Get Claude Team set up properly

We set up Claude Team for small businesses: the seat plan, the security settings, shared Projects and skills written from your actual processes, and the first scheduled routines. If you'd rather have it done once and done right, see how our Claude Team setup works or book a 30-minute call.