Guide · AI tools
Free vs business AI plans: what changes under UK GDPR
ChatGPT, Copilot, Claude, Gemini and Perplexity each treat free and business accounts differently. What each one does with your data, checked October 2026.
By Paulo De Sousa, Co-founder of SimpleKPI3 min read

The same AI tool can be a safe place for client work or a data protection problem, depending on which plan someone is signed in to. Free and personal accounts usually have no contract with your firm and may use what's typed to train the provider's AI. Business plans come with commercial terms and a data processing agreement.
Here is what each of the main providers does, as of 21 October 2026. Providers change these terms often, so check the current terms before you rely on them.
ChatGPT
Free, Go, Plus and Pro are personal plans. OpenAI may use conversations to train its models unless the user turns off "Improve the model for everyone" in their settings. There is no data processing agreement between OpenAI and your firm.
ChatGPT Business (previously called ChatGPT Team) and ChatGPT Enterprise are not used for training by default, and they're the plans to choose for client work.
OpenAI: how your data is used. More detail, including the settings to change: is ChatGPT safe for client data?
Microsoft Copilot
Microsoft Copilot and Copilot Chat signed in with a work account (Microsoft Entra ID) are covered by enterprise data protection. Prompts aren't used to train the underlying models, and Microsoft's data protection addendum applies. One detail worth knowing: when Copilot searches the web, those queries are handled under Microsoft's services agreement rather than the data protection addendum.
Copilot with a personal Microsoft account is a consumer product. Microsoft says it uses Copilot conversations for AI training unless the user opts out of training on their conversation activity, and the countries it excludes don't include the UK.
Microsoft: enterprise data protection. More detail, including work and personal accounts: is Microsoft Copilot safe for client data?
Claude
Claude Free, Pro and Max are consumer plans. Since Anthropic changed its consumer terms in 2025, chats are used for training only if the user allows it, and allowing it means they can be kept for up to five years instead of 30 days.
Claude Team and Claude Enterprise (Claude for Work) and Anthropic's API are under commercial terms and aren't used for training.
Anthropic: updates to consumer terms. More detail: is Claude safe for client data?
Gemini
Gemini with a personal Google account keeps activity by default for adults. Human reviewers may read some conversations, and reviewed conversations can be kept for up to three years. Even with activity turned off, chats are held for up to 72 hours.
Gemini in Google Workspace, signed in with a business account, isn't used to train models without your permission, conversations aren't reviewed by people, and Google's Cloud Data Processing Addendum applies.
Google: Gemini in Workspace privacy. More detail: is Gemini safe for client data?
Perplexity
Perplexity Free, Pro and Max have "AI data retention" switched on by default, which lets Perplexity use your data to improve its models. You can switch it off, but that doesn't remove what was collected before.
Perplexity Enterprise Pro never uses queries for training, and Perplexity acts as a processor under a data processing agreement.
Perplexity: data collection at Perplexity. More detail: is Perplexity safe for client data?
The pattern
| Personal or free plan | Business plan | |
|---|---|---|
| Contract with your firm | No | Yes, with a data processing agreement |
| Used for training | Often by default, or if the user allows it | No |
| Settings controlled by | Each person | Your firm |
The practical rule is simple: approve the business plans your firm pays for, and keep client data off personal accounts. A personal account is fine for questions that involve no personal data at all, such as "explain the difference between a sole trader and a limited company".
Keeping track as plans change
Saiela shows every AI tool by plan, with whether there's a contract, whether it trains on what's typed and where the data goes. You can approve business plans for client work, allow personal accounts for work with no personal data, or block them, and staff on a personal account see a warning before they send. See the AI tools screen.
Not sure which tools and accounts your team is using in the first place? That's often called shadow AI, and our guide on finding shadow AI in a small firm shows you how to find out this week.
General information, not legal advice. Provider terms checked on 21 October 2026.

