The question behind “is our data safe if the team uses ChatGPT?”

Provider terms checked on 22 October 2026.

When a founder or CTO asks whether company data is safe when staff use ChatGPT or Claude, they usually expect a yes or a no. The honest answer is that it depends entirely on which version of the tool your team is using and how it is configured. The same brand name sits on top of three very different commercial deals, and the gap between them is the difference between “your confidential data is protected by contract” and “your confidential data may be retained for years and used to train a model.”

This is general information to help you ask sharper questions, not legal advice. Before you rely on any specific term, confirm the current wording with the provider and your own legal counsel, because these policies change often, sometimes with only a few weeks of notice.

Three doors, three different deals

Both OpenAI and Anthropic sell their models through three broad routes, and each route comes with its own default settings.

  1. Consumer tiers: ChatGPT Free, Plus and Pro, and Claude Free, Pro and Max. Bought by individuals on personal accounts.
  2. Business tiers: ChatGPT Business, Enterprise and Edu, and Claude Team, Enterprise and Education. Bought by an organisation under a business contract. ChatGPT Business is the plan OpenAI sold as ChatGPT Team until it was renamed on 29 August 2025.
  3. The API and developer platform: where systems, not people, send data to the model programmatically.

The single most important thing to understand is that the default answer to “is our data used to train the model?” flips depending on which door you walk through. Get the door wrong and every other control you put in place is decoration.

Consumer tiers: your team may be training the model right now

Here is the uncomfortable part. On the consumer tiers, training is generally on by default.

For ChatGPT Free, Plus and Pro personal accounts, OpenAI says it may use your content to train its models unless you opt out. You switch it off under Settings, Data controls, “Improve the model for everyone.” Even then, giving a thumbs up or thumbs down on a reply means the whole conversation may be used to train its models. OpenAI says deleted chats and temporary chats are removed from its systems within 30 days, unless it is legally required to keep them.

Anthropic made a similar move. In a change announced on 28 August 2025, Claude Free, Pro and Max users (including Claude Code on those plans) were given until 8 October 2025 to choose whether their chats and coding sessions could be used for training. Anthropic’s current wording is that it uses them if you choose to allow it, or if a conversation is flagged for safety review. If you allow it, Anthropic may keep the data in de-identified form for up to five years. If you do not, its existing 30-day retention period applies.

So when a member of staff pastes a client contract, a board pack or a customer list into a personal ChatGPT or Claude account, the realistic assumption is that the content is retained and, unless someone changed a setting or made a choice, may be used to improve the model. That is the scenario that keeps compliance officers awake.

Business tiers: the default flips in your favour

Move to a business contract and the picture changes.

OpenAI states that by default it does not use inputs or outputs from ChatGPT Business, Enterprise, Edu or the API to train its models. Anthropic says the same for its commercial products, such as Claude Team, Enterprise and the API. One exception on Anthropic’s side: if someone uses the thumbs up or down button, it may use that conversation, and it keeps it for up to five years. Both providers offer a Data Processing Agreement for business use, OpenAI’s and Anthropic’s. Both DPAs include the standard contractual clauses with the ICO’s UK Addendum for UK data, and both refer to a sub-processor list.

Business tiers also give you administrative controls. OpenAI says workspace admins control how long data is retained on ChatGPT Business. Anthropic lists single sign-on, domain capture and provisioning for Claude Team. In plain terms, the company owns the account rather than the individual, and an administrator can govern usage.

The catch is that these protections attach to the business product, not to the brand. A member of your team logged into a personal Plus account gets none of them, even though the interface looks almost identical.

The API, zero data retention, and where data physically sits

If you are building AI into a product or a workflow, you are almost certainly using the API, and the rules are different again.

By default, OpenAI generates abuse-monitoring logs for API traffic and keeps them for up to 30 days, unless longer retention is required by law or to protect against harm. It does not train on API data by default. For workloads that cannot tolerate even that, OpenAI offers Zero Data Retention (ZDR) and Modified Abuse Monitoring, both subject to prior approval. Under ZDR, eligible customer content is excluded from those logs.

Anthropic deletes API inputs and outputs within 30 days by default, with exceptions such as a zero data retention agreement, usage policy enforcement and the law. Some Claude API and Claude Code Enterprise customers can, subject to approval, get an arrangement where Anthropic does not store their inputs or outputs. Worth knowing: even then, Anthropic still keeps the results of its safety classifiers, so “zero retention” is not literally zero for every byte.

Data residency matters too, and it differs sharply between the two. OpenAI offers UK storage at rest for eligible API customers and for new ChatGPT Enterprise and Edu customers, announced on 22 October 2025. That page lists ChatGPT Business as not eligible, and it separates storage at rest from where the model runs: its inference residency regions do not include the UK. Anthropic’s data residency page lists the US as the only workspace geography. If your data must stay in a particular region, treat residency as a contract term to confirm in writing, not an assumption.

Retention is not the same as deletion

The most important lesson of the last year is that a stated retention policy can be overridden by a court.

In the New York Times copyright case against OpenAI, a court order required OpenAI to keep output logs that it would normally have deleted, including chats users had deleted. OpenAI says the order covered ChatGPT Free, Plus, Pro and Team (now Business) and API traffic without a ZDR agreement. It also says the order did not cover ChatGPT Enterprise, Edu or ZDR API customers. OpenAI says its obligation to keep new data ended on 26 September 2025, that it returned to deleting removed chats within about 30 days, and that it still stores limited April to September 2025 data. In January 2026 a US district judge upheld orders requiring OpenAI to produce a sample of 20 million de-identified ChatGPT logs.

Two takeaways for a leader. First, “we delete after 30 days” is a default, not a guarantee, because litigation and legal holds can freeze it. Second, and more reassuring, the plans OpenAI says sat outside the order were its enterprise and ZDR ones. That is one more concrete reason to put sensitive work on a business or ZDR contract rather than a consumer login. Confirm the current position, because the case is still moving.

Shadow AI: the paste is the risk, not the tool

Notice that almost every risk above comes from the same act: a person pasting company data into a consumer account. This is “shadow AI,” the unsanctioned use of AI tools that IT never approved and cannot see. People reach for whatever makes the work easier, whether or not anyone approved it.

Banning AI does not fix this. It drives the behaviour underground, because the productivity pull is real and people will use their phones if they have to. The fix is to make the safe path the easy path: give people an approved, business-tier tool that is at least as good as the one they would otherwise reach for, and be explicit about what may and may not go into it.

What this means for a UK firm of 5 to 50 people

If you use a business plan, your firm is normally the controller and the provider is your processor. The ICO’s guidance on controllers and processors explains the split: you decide why and how personal data is used, and the provider handles it on your instructions. That is why the signed DPA matters. The ICO’s guidance on AI and data protection sets out what controllers should consider when they use AI. The ICO says it is under review because of the Data (Use and Access) Act 2025, so check it for updates.

Sending UK personal data to a provider that processes it outside the UK is a restricted transfer, and you need a safeguard, as the ICO’s international transfers guidance explains. The UK-US data bridge covers US companies certified on the Data Privacy Framework list. When we searched that list by name on 22 October 2026, it returned no entry for OpenAI or for Anthropic, so we would not rely on the bridge for either. A group company could be listed under another name, so check the list yourself. Both DPAs instead use the standard contractual clauses with the ICO’s UK Addendum: see OpenAI’s DPA section 4.2 and Anthropic’s DPA Schedule 3.

A firm of 10 will usually be on ChatGPT Business or Claude Team, not Enterprise. On those plans you get the no-training default and the option to sign a DPA. You do not get a UK storage option: OpenAI’s UK offer lists API and new Enterprise and Edu customers, and Anthropic lists only the US. For most small firms that is a risk to weigh, not a reason to ban the tools. Redact what you send, and move the sensitive workloads to a contract that can offer more.

Checked on 22 October 2026. Each cell links to the provider’s own page.

OpenAITrains by defaultRetentionUK storage optionSource
Consumer (Free, Plus, Pro)Yes, unless you opt outDeleted chats removed within 30 days, unless legally required to keepNone listedModel training, retention
Business (ChatGPT Business)NoAdmins set retention; deleted chats removed within 30 days, unless law or harm prevention requires longerNot on the eligibility listEnterprise privacy, residency
APINoAbuse logs up to 30 days; ZDR on approvalUK storage at rest for eligible customersAPI data controls
AnthropicTrains by defaultRetentionUK storage optionSource
Consumer (Free, Pro, Max)No: only if you choose to allow itDeleted chats removed within 30 days; up to 5 years in de-identified form if you allow trainingNone listedModel training, retention
Business (Claude Team)NoDeleted chats removed within 30 daysNone listed; the US is the only workspace geographyModel training, retention
APINoDeleted within 30 days; ZDR on approvalNone listed; the US is the only workspace geographyRetention, residency

A safe-rollout checklist

You can close most of the exposure with a short, practical programme, and an AI readiness audit is a fast way to see where your current exposure sits. Work through these roughly in order.

  1. Publish an approved-tools list. Name the specific tiers people may use (for example, your ChatGPT Business or Claude Team workspace) and state plainly that personal Free, Plus, Pro and Max accounts are not approved for company data. If you have no written policy yet, our free one-page AI policy puts the list and the rules on one sheet. The small business AI policy template goes further.
  2. Buy the business tier and put everyone on it. The default no-training protection, DPA and admin controls only apply to the business contract, so this is the most useful single step. Our guide to rolling out ChatGPT or Claude to your team covers it, and Copilot vs ChatGPT vs Claude helps you choose where each fits.
  3. Sign and file the DPA. Confirm it covers GDPR Article 28, international-transfer clauses and the sub-processor list, and store it where your auditors can find it.
  4. Enforce SSO and provisioning where your plan offers them. Route access through your identity provider so joiners and leavers are handled automatically and usage is attributable to a named person.
  5. Turn off training and set retention deliberately. Even on business tiers, check the toggles and configure retention to match your data-classification policy rather than accepting whatever the defaults happen to be.
  6. Decide whether you need ZDR or data residency. For regulated or highly sensitive workloads, pursue a zero-data-retention agreement and confirm the processing region in writing.
  7. Redact and minimise at the point of entry. Train people, and better still build tooling, to strip names, card numbers and secrets before data reaches the model. The safest data is the data you never send.
  8. Classify your data first. None of the above works if nobody knows which documents are confidential, so a simple traffic-light classification usually has to come before everything else.
  9. Review quarterly. These terms change, so re-check provider policies and your own settings on a schedule rather than assuming last year’s answer still holds.

The safe path is an engineering decision

Treat all of this as an engineering choice rather than a policy memo. The controls that keep data safe (the right tier, no-training settings, a signed DPA, redaction at the point of entry) are the same controls that make an AI system reliable and auditable. It is the mindset behind how we build at Rogue. The enquiry auto-reply behind speed-to-lead on our own site is a fixed template, and no AI model reads the enquiry. Our own prospecting engine runs every draft through an adversarial verifier, a separate model acting as a judge, and a person reads and sends every message, so nothing leaves the system unchecked. Whether you buy AI off the shelf or build it into your workflows, the principle holds: know which door your team is using, make the safe door the default, and verify the settings rather than trusting the brand name on the box.