If I Turn Off AI Training on Our Data, Can I Enter Confidential Information?

No, not based on that setting alone. Turning off AI training is a good step, but there is more to the story.

Depending on the provider, product, and account settings, turning off training generally means eligible new content, such as prompts, uploaded files, or chats, will not be used to improve the vendor’s models. It lowers one important risk: your information being reused to improve the AI model.

But training is only one part of the confidentiality, privacy, and security picture. You still need to know where the information is processed, how long it is kept, who can access it, whether it is reviewed by people, how backups are handled, and whether other connected apps or plug-ins can receive it.

A business or enterprise account with the AI provider, rather than a consumer account, can provide stronger protections. It may include a contract with your organization, no training by default, admin controls, audit logs, and retention and deletion controls. Those protections are helpful, but they do not constitute automatic approval. Someone still needs to review the actual terms, settings, data flows, and intended use.

The simple rule is this: do not enter confidential, personal, or other sensitive information into a personal consumer AI account or any other account outside your organization’s approved, managed environment. Use such information only in an organization-approved and organization-managed service that has been assessed and authorized for that type of information and that specific purpose.

Even then, share only what is needed. Minimize the information and remove or generalize identifying details whenever you can. Do not assume that removing names alone makes information anonymous or non-confidential. Some information, including client details, pricing, health information, or financial information, may still be off limits because of company policy, privacy law, contracts, or professional duties.