ChatGPT Business vs free and personal AI accounts: what changes

#ChatGPT Plus vs ChatGPT Business

Plus is a subscription for one person. Business is a shared workspace for an organisation. For your data, they differ on five points.

Training is the difference most people know. On Free, Go, Plus and Pro, OpenAI may use your chats to train its models until you turn off "Improve the model for everyone" under Settings > Data controls, or choose "Do not train on my content" in the Privacy Portal. One of the two is enough, and the steps are in our guide on how to turn off AI training on your data. Even after opting out, a thumbs up or down can send the whole conversation to training. On Business, OpenAI does not train on workspace data by default, and feedback does not change that (OpenAI, model training).

For an organisation, the other rows carry more weight. OpenAI signs a data processing addendum for Business, Enterprise and the API (OpenAI, enterprise privacy), which is what Article 28 of the GDPR asks for when a vendor processes personal data on your behalf. Single sign-on and domain verification are part of Business, so members sign in through your own identity provider (OpenAI, SSO for Business). A workspace needs at least 2 seats (OpenAI, Business overview); current prices are on OpenAI's pricing page.

Enterprise adds what larger organisations ask for: SCIM provisioning from the user directory, data residency and the Compliance API, which gives admins an audit log of conversations. Memory, shared links and connected apps are routes of their own, covered in ChatGPT privacy risks at work.

#The four vendors side by side

For the four tools most teams reach for, the gap between tiers looks like this. The vendor sources were checked on 20 October 2026.

A few rows need context.

Google draws the sharpest consumer-versus-business line. In the Gemini app on a personal account, chats are used to improve Google's AI while Keep Activity is on, a subset is read by human reviewers and reviewed chats are kept for up to three years (Gemini Apps Privacy Hub). With a Workspace account, your content is not reviewed by humans or used for model training outside your organisation without permission, and it falls under Google's Cloud Data Processing Addendum (Workspace privacy hub). The two versions are compared in detail in Gemini privacy risks at work.

Microsoft is just as clear. In Microsoft 365 Copilot, prompts, responses and data accessed through Microsoft Graph are not used to train foundation models, the service falls under Microsoft's Data Protection Addendum and EU traffic stays inside the EU Data Boundary, except when Anthropic models are used (Microsoft Learn). Microsoft now calls the work version simply Microsoft Copilot, the same name as the consumer app, so the account you sign in with is what tells them apart. On a personal account, it depends on the version of the Copilot app. The older app uses conversations for training unless you opt out (Microsoft, Copilot privacy FAQ). For the app updated on 18 August 2026, Microsoft states that prompts, responses and file contents are not used to train foundation models (Microsoft, Copilot activity history).

Anthropic changed its consumer terms on 28 August 2025. Since then, Free, Pro and Max chats, Claude Code sessions included, are used for training when the user's training setting is on, and are then kept for up to five years instead of 30 days. Existing users had to choose by 8 October 2025. Team, Enterprise, education, government and API use fall under Anthropic's Commercial Terms and were not affected (Anthropic, updates to consumer terms). Those terms include Anthropic's data processing addendum, and the Team plan includes single sign-on (Claude pricing). The plan-by-plan detail is in Claude privacy risks.

The pattern holds: the four vendors differ in the details, but the dividing line falls between consumer and business every time, not between one brand and another.

#Can your admin read your chats?

In ChatGPT Business other members cannot see your chats, but workspace admins can view, export and delete them (OpenAI, enterprise privacy). "No training" does not change that, so assume your admin can read a work chat. For Gemini, Copilot and Claude, and for what the GDPR allows, see our article on what your employer can see, tool by tool.

#The same question, beyond the four names everyone checks

ChatGPT, Gemini, Copilot and Claude are the four a leader thinks to ask about, because they are the four with a marketing budget. They are rarely the only four in use. BeeSensible's catalogue runs this same tier comparison, training default and all, against more than 1,000 AI tools, and shows it for the AI tools your team actually opens, including the ones nobody thought to ask about.

#Why this is a leadership decision

A free account is a choice made by an individual. A business licence is a choice made by the organisation. That is exactly where the governance question sits.

While teams run on free accounts, nobody has a central view of which settings are on, which shared links are circulating or which data has landed in which history. There is no data processing agreement to fall back on when a regulator asks under what terms the tool was running. And when someone leaves, their access walks out the door with everything in it.

Menlo Security's 2025 report shows how ordinary this is: 68 per cent of employees use free AI tools such as ChatGPT through a personal account, and 57 per cent enter sensitive data (Menlo Security, August 2025). That is not recklessness. It is a free account that was already open when the work arrived.

So choosing the right tier is not an IT detail. It decides whether the organisation has any legal and governance footing at all for the AI use that is already happening.

#How to choose as a leader

The decision comes down to a few questions.

Does the team work regularly with personal data, customer records or confidential context? Then you need a managed business or enterprise tier, with a data processing agreement, training excluded and central control. A personal account is not an approved environment for that work.

Do you need SCIM provisioning or a Compliance API for audit and eDiscovery? At OpenAI and Anthropic, those come with Enterprise. Single sign-on is already part of ChatGPT Business and Claude Team.

Is the team still on free accounts today? The first step is not a ban but moving the work to a managed tier. A ban on its own just pushes use to accounts you cannot see. Record in an AI register which variant of each tool is allowed and under what condition, so the decision does not live only in an email thread.

For the legal footing, it helps to have your GDPR measures for AI use and the data processing agreement for each AI tool lined up before you roll a tool out broadly.

#What the right tier still does not solve

Say the decision is made. The team runs on a managed enterprise tier, training is off, the data processing agreement is signed and the admin has control. Most of the account-level risk is now covered.

One thing remains. The account tier governs what happens to data after it reaches the tool. It does not decide whether the employee should have pasted the account number, the card number or the health note in the first place.

A data processing agreement does not redact a prompt. A training exclusion does not notice that a screenshot shows a customer name in the browser tab. An enterprise tier does not stop someone sending a full HR file for a wording suggestion. The data still reaches the model; the difference is that it will not be used for training and falls under a DPA. That is real, but it is not the same as sharing less sensitive data.

What is left is the human moment: the few seconds in which the employee can still see what is sensitive and mask or remove it.

#Spot sensitive details before the prompt is sent

BeeSensible highlights sensitive data in text fields while the employee is still typing. The text is sent to BeeSensible's EU detection service, where analysis runs in memory and the text is discarded straight after detection. In ChatGPT, Gemini, Copilot and Claude, names, email addresses, phone numbers, account numbers, card numbers and other configured categories get a highlight before anyone presses Enter.

The user decides: mask it with a placeholder, or remove it, unless the organisation has set masking to happen automatically when text is pasted or sent. BeeSensible does not store the prompt text. Admins see patterns by category and application, without reading employee chats. And if the organisation has allowed the business variant of a tool, the notice on opening can point out that the account is not recognised as a business account.

This works on any account tier, including the right one. A managed enterprise tier and a signed data processing agreement set the terms; the highlight in the text field deals with the paste itself.

The account tier is the biggest lever a leader can pull. Set it to business, with a DPA and training excluded. Then cover the gap that is left, where AI data leakage usually begins: the ordinary paste, just before the prompt is sent.