AI assistants can process more than the text you type. Depending on the service and features you use, data may include your prompts and attachments, the assistant’s replies, technical and usage information, and context from your device or connected apps. Providers say they use this information to answer requests, operate and improve their services, personalize responses, and protect users. What is collected, retained, or used to train models varies by provider, account type, settings, app version, and region.
What data can an AI assistant collect?
The precise scope depends on the assistant and the features you enable. A text-only exchange differs from a voice conversation, an image prompt, or a request that uses browser content or a connected app. The examples below describe disclosures from OpenAI, Google, Anthropic, and Microsoft; they do not establish what every AI assistant collects.
As an Amazon Associate I earn from qualifying purchases.
Information you submit
Prompts, conversation content, feedback, and files or images you provide are common categories in provider disclosures. Google’s Gemini Apps Privacy Notice, dated June 29, 2026, describes a wider range of feature inputs, including videos, photos, browser-page content, imported chats, Gemini Live transcripts and recordings, and instructions supplied through customization features. Those categories apply when the relevant feature or input is used; they should not be read as a claim that every session captures all of them.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsInformation created or observed while the service runs
Some disclosures also cover the assistant’s generated output and technical or usage information. Google’s notice lists app, browser, and device information; interaction logs; performance metrics; crash and debugging information; system permissions; some device context; and information from connected apps. Which of these is relevant can depend on the feature, permissions, and session.
#1 Best Overall
Google also says Gemini may use location to make responses relevant. Its notice distinguishes precise device location, which requires permission, from a general area inferred from an IP address or account addresses; location stored with Gemini activity is coarsened to a general area. This is a Google-specific disclosure, not a rule for all assistants.
How do providers say they use the data?
Provider policies describe several purposes, and answering a prompt is only one of them. Google lists providing, maintaining, improving, and developing services; personalization and customization; communication; performance measurement; and protecting Google, users, and the public. OpenAI describes model improvement for eligible consumer content subject to settings, as well as limited content access for safety and security investigations, support, legal matters, or improvement unless a user opts out. Anthropic’s consumer policy includes improvement under specified conditions and safety review.
These are providers’ stated purposes, not independent verification of how each internal process operates. A feature may also send relevant context—such as content from a file or connected app—to the assistant so it can respond to the request.
Does collecting data mean it will train the model?
No. Service operation, history or activity storage, personalization, safety handling, human review, and foundation-model training are distinct uses. Turning off a training setting does not necessarily stop the service from processing a prompt to answer it, erase saved history, or remove all retention. Provider-specific examples published as of October 4, 2026 include:
| Service and scope | Training and related controls described by the provider |
|---|---|
| ChatGPT consumer accounts | OpenAI says turning off “Improve the model for everyone” excludes new conversations from model training; saved chats remain in history. Temporary Chats do not appear in history, update memories, or improve models, but OpenAI says they may be retained for up to 30 days for safety. |
| Google Gemini Apps | Google says that with Keep Activity off, chats are still used to respond and help protect users, including through human reviewers. The setting therefore should not be interpreted as “nothing is processed.” |
| Claude consumer accounts | Anthropic says chats and coding sessions may be used to improve models if the user allows it, if conversations are flagged for safety review, or through another explicit opt-in. Incognito chats are not used to improve Claude, even when Model Improvement is enabled. |
| Claude commercial products and API | Anthropic says commercial product inputs and outputs are not used for training by default; feedback or explicit permission may allow use. Commercial and API terms differ from consumer rules, and zero-data-retention arrangements have eligibility and feature limitations. |
| Updated Microsoft Copilot app for individuals | Microsoft says prompts, responses, and file contents in the updated app are not used to train foundation models. Optional feedback may be used to improve Copilot, but not to train those foundation models. This disclosure is specific to the updated app and does not automatically describe work or school accounts. |
The Microsoft distinction matters because Microsoft’s updated Copilot app became available on August 18, 2026, and Microsoft warns that some older-app guidance applies only to the legacy version. Check that a setting or policy refers to the app and account you actually use.
How long is assistant data retained?
There is no single retention period for “AI assistant data.” Different rules can apply to ordinary conversations, temporary sessions, activity records, feedback, safety-review material, or API logs. These examples are provider-specific policy statements current in 2026, not a market-wide standard.
Rank #4
| Provider and data type | Published retention information |
|---|---|
| OpenAI ChatGPT Temporary Chats | OpenAI says they may be kept for up to 30 days for safety. |
| Google Gemini Apps Activity | Google lists auto-delete choices of 3 months, 18 months (the default), 36 months, or indefinite retention. |
| Google Gemini data reviewed by human reviewers | Google says chats and related data reviewed by human reviewers may be retained for up to three years. Deleting Gemini activity does not delete this reviewer-held data. |
| Anthropic consumer feedback | Anthropic says feedback may result in retention of the entire related conversation for up to five years. |
| OpenAI API abuse-monitoring logs | OpenAI says these logs may contain prompts and responses and are retained for up to 30 days by default, subject to exceptions; eligible customers may be able to access other controls. |
For a particular account, consult the current policy for that service and region. A deletion action may apply to one data store or category without deleting copies governed by a separate safety, feedback, connected-service, or organizational rule.
Do personal, work, and API accounts follow the same rules?
Usually, the relevant terms depend on the account and product, not just the assistant’s brand name. Consumer-app settings should not be assumed to govern an employer’s workspace or a developer’s API use.
Best Value
- OpenAI workspaces: OpenAI says Business, Enterprise, Edu, and Healthcare workspace content is not used for model training by default. Organization-level retention and access settings still matter.
- OpenAI API: API inputs are not used for training by default, according to OpenAI. Abuse-monitoring logs may contain prompts and responses and are retained for up to 30 days by default, subject to exceptions; approved eligible customers may have other controls.
- Anthropic: Commercial and API rules differ from Claude Free, Pro, and Max consumer rules. The provider’s zero-data-retention arrangement is not universal and has eligibility and feature limitations.
- Microsoft: Individual Copilot disclosures do not automatically cover work or school accounts using Entra ID or Microsoft 365 contexts.
For sensitive work, the organization’s contract, administrator settings, and retention policy are more relevant than assumptions based on a personal account’s controls.
What should you check before sharing information?
- Identify the exact service and account. Confirm the product or app version, region, and whether you are signed in with a personal, work or school, or API account.
- Separate training controls from storage controls. Check model-improvement settings independently from chat history or activity settings.
- Review memory and personalization. Removing a saved memory may not delete the conversation that supplied it.
- Check feature permissions. Before enabling voice, camera, location, screen, browser, or connected-app features, review what access each one requires and what context it can use.
- Understand deletion scope. Find out whether deletion affects only visible history or also activity records, reviewer-held copies, connected-service records, or backups.
- Use the right terms for workplace data. Check the organization’s contract and administrator policy rather than relying on consumer-app disclosures.
- Minimize sensitive submissions. Avoid sharing passwords, credentials, confidential material, or sensitive personal data unless the relevant service and account terms have been reviewed and the sharing is appropriate. OpenAI and Google caution users about entering sensitive or confidential information.
How to compare assistants fairly
Compare like with like: a consumer app against another consumer app, or business/API offerings against the same type of offering. For each one, check the same questions:
- What can the user submit—text, files, images, audio, video, screens, or feedback?
- What context may be collected through the app, browser, device, location, permissions, or connected services?
- Which purposes are stated: responding, service improvement, foundation-model training, personalization, safety, or human review?
- What separate controls exist for training, history or activity, memory, temporary or incognito sessions, and permissions?
- What retention and deletion rules apply to ordinary chats, temporary chats, feedback, safety-review material, and API logs?
- Does the policy match the account type, app version, region, and workspace or administrator settings in use?
The examples here cover official disclosures for OpenAI/ChatGPT, Google/Gemini, Anthropic/Claude, and Microsoft/Copilot. They are not a complete market comparison, and policies can change; consult each provider’s current privacy and product documentation before relying on a setting or retention period.
Free tools Windows power users keep installed
One-click scans. No signup required.
Quick Recap
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.
