Recommended Free Tools
For confidential work, the safer starting point is an organization-managed account that your employer has approved—not a personal chatbot account with a training setting switched off. ChatGPT Business or Enterprise, Microsoft Copilot Chat with a work or school account, Claude under an organizational agreement, and Gemini in a qualifying Google Workspace edition each describe protections for certain business uses. They differ in retention, administrator access, connected tools, and coverage, so none is automatically safe for every task or compliant with every law.
Before sharing sensitive information, confirm the exact product, account, plan, configuration, and contract your organization permits. “Not used for training” does not mean “never stored,” “invisible to administrators,” or “compliant with our requirements.”
What makes a chatbot safer for sensitive work?
Safety depends on the whole service and its setup, not just the chatbot brand. Check how the product treats prompts and responses, how long it retains them, who in your organization can access them, and what happens when it searches the web or connects to company data.
- Account and terms: Is this a managed organizational account covered by your employer’s agreement, or a personal account with consumer terms?
- Training: Are prompts and outputs excluded from model training by default, or only after a setting is changed?
- Retention and deletion: Are records stored, for how long, and do different features or surfaces have different retention rules?
- Internal access: Can authorized administrators review conversations through audit, compliance, search, or retention tools?
- Connected services: Does the assistant access company files, other apps, or the public web? Those pathways may have different permissions and terms.
- Scope: Do the stated protections apply to your edition, region, product surface, organization, and configuration?
A training exclusion addresses one use of data. It does not, on its own, answer the other questions.
Free tools Windows power users keep installed
One-click scans. No signup required.
#1 Best Overall
How the managed options compare
The table summarizes vendor-described protections and qualifications. “Managed” does not mean that every feature is enabled or that every account receives the same terms; verify the applicable agreement and settings with your organization.
| Service and account context | Training use | Retention and internal access | Important scope checks |
|---|---|---|---|
| ChatGPT Business, Enterprise, Edu, Healthcare, Teachers, or API | OpenAI says inputs and outputs are not used for training by default on these services. | OpenAI describes retention controls for qualifying organizations and zero data retention for eligible API customers. It also lists access controls and compliance features; administrator access depends on organizational settings and applicable features. | Confirm product, eligibility, retention configuration, and region. OpenAI says data residency at rest in named regions is available to eligible Enterprise, Edu, Healthcare, and API customers, with feature coverage depending on eligibility. OpenAI’s security page reports SOC 2 Type 2 examination coverage for relevant API and ChatGPT business service controls and lists ISO certifications for some services; these do not establish compliance for a customer’s particular workflow. |
| Microsoft Copilot Chat signed in with a work or school account | Microsoft says prompts and responses are not used to train foundation models under enterprise data protection. | Microsoft says prompts, Bing search queries triggered by prompts, and responses are logged. IT administrators can use Microsoft search and audit tools to view this information. | Protection is tied to a work or school identity. Subscription affects available controls. Microsoft’s documentation distinguishes web search queries from the main prompt-and-response path and Microsoft Graph; check the applicable terms and settings. |
| Claude Platform API with an organization-level zero-data-retention arrangement | The cited retention documentation focuses on storage and does not establish a general training-use claim for every Claude product or agreement; check the terms for the exact service. | Anthropic says that, under an enabled zero-data-retention arrangement, API prompts and responses are not stored at rest after the response returns. Coverage is surface-specific: claude.ai chats, files, and projects follow the organization’s retention policy unless deleted earlier, while Activity Feed and some session transcripts may have longer retention. Some metrics logging may fall outside the arrangement. | The arrangement must be requested and enabled separately for each organization. Verify the exact product, organization, model, contract, and surface. |
| Gemini under a qualifying Google Workspace edition | Google says it will not use customer data to train or fine-tune supporting generative AI models without the customer’s prior permission or instruction. | Google says customer data is processed under the Workspace agreement and that existing Workspace security and data controls apply. The specific handling depends on the Workspace product used. | Confirm that the edition and feature qualify for Workspace protections. Gemini Notebook makes a separate copy of Drive sources in Notebook data; the organization’s file-sharing and data-region settings do not apply to that copy. Google’s Workspace Generative AI Privacy Hub was last updated August 14, 2026. |
What changes between personal and work accounts?
ChatGPT
OpenAI’s business-service commitments are distinct from consumer account controls. For ChatGPT Temporary Chats, OpenAI says chats do not appear in history, do not create or update memories, and are not used to improve models. They may still be retained for up to 30 days for safety. Saved chats, memory, and training controls have separate behavior, so Temporary Chat is not a substitute for an approved business account or a guarantee of no retention.
Rank #2
Microsoft Copilot
Microsoft distinguishes consumer Copilot from Copilot Chat used with a work or school account. Its personal-account support guidance says users can opt out of using future conversations for model training, but that choice does not exclude conversations from other product or system improvement, advertising, safety, security, and compliance uses. That support article says it covers an older app version after an updated app became available August 18, 2026, and explicitly directs work or school users to enterprise data protection.
Claude and Gemini
For Claude, an organization-level API retention arrangement should not be assumed to cover claude.ai chats or every other product surface. For Gemini, the Workspace protections described by Google apply to qualifying Workspace use; they should not be presumed to cover a personal Google account or every feature in the same way.
How to choose and use an option responsibly
- Ask your organization which exact service is approved. Confirm the account identity, product name, edition, and whether the work is permitted under your company’s policy. A personal account is not made approved simply by turning off training.
- Match the task to the data rule. Determine whether the material is public, internal, confidential, regulated, or contract-restricted. If policy bars sending a category of data to an external service, do not paste it into a chatbot without explicit authorization.
- Verify training and retention separately. Read the terms for prompts and outputs, then check storage, deletion, retention schedules, and any exceptions for specific features. Ask whether a claimed zero-retention arrangement actually covers the surface you will use.
- Check who can review activity. Ask your administrator what audit, search, compliance, and retention systems can expose, and which authorized roles can use them. Workplace protections can coexist with organizational visibility.
- Inspect connected data and web search. If the assistant can retrieve company files or search the web, confirm permissions and how those requests are handled. Microsoft, for example, describes separate handling for Bing web queries; Google identifies a distinct copy and data-region caveat for Gemini Notebook sources.
- Confirm region, contract, and configuration. For regulated or contract-restricted work, have privacy or security staff verify the exact applicable agreement, processing geography, feature eligibility, and settings. Vendor certifications and broad product statements do not certify your particular use.
- Minimize what you submit. When use is approved, remove names, account numbers, secrets, and unnecessary source material. Share only the minimum information needed for the task, and avoid credentials, access tokens, or authentication codes.
Questions to take to your administrator
- Which chatbot product and account type are approved for this data classification?
- Does the applicable agreement exclude prompts and outputs from training by default, and are there exceptions?
- What retention and deletion rules apply to chats, files, search queries, logs, and connected features?
- Can administrators or compliance personnel retrieve conversations, and under what policies?
- Are web search, company-file retrieval, or third-party integrations enabled, and do they follow different data terms?
- Are the relevant features covered in our plan and region, and has the required configuration been applied?
If the answer to a material question is unclear, treat the service or feature as unapproved for that sensitive content until your organization confirms its use.
Quick Recap
Rank #4
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
