The safest rule is simple: send an AI app only information you would accept being processed under that app’s specific terms. Privacy toggles can limit training, history, memory, personalization, or connected-app access, but they do not turn a consumer service into an offline, private tool. Before each sensitive prompt, identify the exact app and account, minimize the data, check the relevant controls, and understand what may still be retained or reviewed.
Start with data minimization
Privacy protection begins before you open a settings page. The less identifying information an app receives, the less it can expose, retain, or associate with you if its policy, account access, or security conditions change.
As an Amazon Associate I earn from qualifying purchases.
Remove direct identifiers
Replace names with neutral labels such as Client A; remove account numbers, addresses, telephone numbers, email addresses, precise dates of birth, employee IDs, and customer references. If an answer depends on age, region, or role, provide only the minimum needed—for example, “a 42-year-old patient in Ontario” rather than a full name and medical record.
Redaction reduces practical risk; it is not a guarantee of anonymity. A combination of dates, job titles, locations, unusual events, and quoted text can still identify a person. Rewrite distinctive passages and remove metadata from files when it is not needed.
#1 Best Overall
- POWERFUL SECURITY KEY: The Security Key C NFC is the essential physical passkey for protecting your digital life from phishing attacks. It ensures only you can access your accounts.
- WORKS WITH 1000+ ACCOUNTS: Compatible with Google, Microsoft, and Apple. A single Security Key C NFC secures 100 of your favorite accounts, including email, password managers, and more.
- FAST & CONVENIENT LOGIN: Plug in your Security Key C NFC via USB-C and tap it, or tap it against your phone (NFC) to authenticate. No batteries, no internet connection, and no extra fees required.
- TRUSTED PASSKEY TECHNOLOGY: Uses the latest passkey standards (FIDO2/WebAuthn & FIDO U2F) but does not support One-Time Passwords. For complex needs, check out the YubiKey 5 Series.
- BUILT TO LAST: Made from tough, waterproof, and crush-resistant materials. Manufactured in Sweden and programmed in the USA with the highest security standards.
Use excerpts or synthetic examples
Do not upload an entire contract, case file, spreadsheet, meeting recording, or source-code repository when a short excerpt answers the question. Create a fictional example with the same structure, column types, or error pattern. Summarize confidential material yourself and ask the app to work on the summary.
For documents, inspect what is included beyond the visible page: comments, tracked changes, hidden sheets, revision history, embedded images, and author metadata. For audio, video, and screen sharing, remember that the service may receive everything in the captured frame or recording, not just the item you intended to discuss.
How do I protect my privacy when using AI apps?
Use this sequence for every app and every account. Labels and locations change, so verify the current interface and privacy notice immediately before using a new feature.
- Classify the material. Decide whether the prompt contains credentials, financial information, health information, legally protected material, trade secrets, customer data, or another person’s personal information.
- Minimize it. Delete identifiers, shorten the text, and substitute a synthetic example where possible.
- Identify the product and account. Confirm whether you are in a personal consumer app, a work or school deployment, or an API or managed service. Similar names do not mean identical terms.
- Check improvement or training controls. Find the setting that governs use of content to improve models. Determine whether it applies only to future conversations and whether feedback, uploads, or other features are excluded.
- Check history and activity. Review where chats are stored, the default retention period, auto-delete choices, and whether deleting a conversation also removes saved activity or memory.
- Review memory and personalization. Remove information already saved and disable personalization you do not need. A training opt-out does not necessarily erase memory.
- Audit connected apps and permissions. Disconnect integrations that are unnecessary and read the connected service’s own terms. Turning off an AI setting cannot control data retained by another service.
- Choose a temporary mode for a one-off task when appropriate. Read its retention and safety exceptions first; “temporary” does not automatically mean immediate deletion or zero human review.
- Delete what you can afterward. Remove the chat, uploaded file, saved memory, and connected-app permission separately when the product provides separate controls.
Check the account type before trusting a privacy promise
Consumer, workplace, school, and API products can process the same prompt under different rules. Never transfer a promise made for a managed business service to a personal account.
Personal consumer accounts
These commonly allow the provider to use conversation content for service improvement unless you change a control. The setting may be account-wide, product-specific, or limited to new conversations. A subscription or paid individual plan does not by itself make content confidential.
Work or school accounts
An organization may negotiate different data protections, while administrators can retain or inspect logs. Your employer’s policy, discovery obligations, and security monitoring still apply even if prompts are excluded from foundation-model training.
API and managed services
API data handling is governed by the API agreement and configuration, not by the consumer app’s controls. Confirm retention, abuse monitoring, regional processing, logging, and who can access the account before sending production data.
Recommended Free Tools
Rank #2
- POWERFUL SECURITY KEY: The YubiKey 5C NFC is the most versatile physical passkey, protecting your digital life from phishing attacks. It ensures only you can access your accounts
- WORKS WITH 1000+ ACCOUNTS: Compatible with popular accounts like Google, Microsoft, and Apple. A single YubiKey 5C NFC secures 100+ of your favorite accounts, including email, password managers, and more
- FAST & CONVENIENT LOGIN: Plug in your YubiKey 5C NFC via USB and tap it, or tap it against your phone (NFC), to authenticate. No batteries, no internet connection, and no extra fees required
- MOST SECURE PASSKEY: Supports FIDO2/WebAuthn, FIDO U2F, Yubico OTP, OATH-TOTP/HOTP, Smart card (PIV), and OpenPGP. That means it’s versatile, working almost anywhere you need it
- PRIMARY & SPARE KEYS: Just like having a spare house key, we recommend buying two YubiKeys - one for daily use and one as a spare. That way you’ll never get locked out of your accounts
Can AI apps use my chats to train their models?
Sometimes. “Training,” “model improvement,” “personalization,” “safety review,” and “service operation” can be separate uses. An opt-out usually addresses future eligible content; it does not promise that every copy is erased, that prior human review is undone, or that feedback is excluded.
Read the scope next to the toggle. Check whether it covers uploaded files, voice and video, screen content, memories, imported chats, connected applications, and thumbs-up or thumbs-down feedback. If the notice is unclear, treat the material as shareable only if you are comfortable with the broader processing described for that account.
ChatGPT: stopping model-improvement use on individual plans
OpenAI says content from Free, Plus, and Pro individual services may be used to train models. Its documented control is Settings > Data controls > Improve the model for everyone; users can also use the Privacy Portal. Menu names may change, so confirm the current label in your account.
What the individual opt-out does
OpenAI says turning the control off excludes new conversations from model improvement. It is not a promise that older conversations, security logs, or other operational records disappear. If you submit response feedback, OpenAI says the conversation associated with that feedback may be used for model improvement even after you opted out. Do not send feedback containing details you would not want processed under that exception.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Clear out junk files and repair common Windows errors3Fix the driver behind crashes, sound loss and screen glitchesTemporary chats
OpenAI says temporary chats do not appear in chat history, do not create or update memories, and are not used to improve models while they are temporary. Those properties make the mode useful for a one-off question, but they do not make the service equivalent to an offline tool or establish that no operational or safety processing occurs.
Business, Enterprise, Edu, and API inputs
OpenAI states that inputs and outputs for Business, Enterprise, Edu, and API offerings are not used for model improvement by default. Verify that you are actually authenticated to that offering, that your administrator has not changed a configuration, and that files or connected services fall under the same agreement.
Claude: consumer controls and the Incognito exception
Anthropic’s consumer help article, dated May 22, 2026, says users can change model-improvement settings. Anthropic also says Incognito chats are not used to improve Claude even when the main setting permits improvement.
Rank #3
- POWERFUL SECURITY KEY: The YubiKey 5 NFC is the most versatile physical passkey, protecting your digital life from phishing attacks. It ensures only you can access your accounts
- WORKS WITH 1000+ ACCOUNTS: Compatible with popular accounts like Google, Microsoft, and Apple. A single YubiKey 5 NFC secures 100+ of your favorite accounts, including email, password managers, and more
- FAST & CONVENIENT LOGIN: Plug in your YubiKey 5 NFC via USB and tap it, or tap it against your phone (NFC), to authenticate. No batteries, no internet connection, and no extra fees required
- MOST SECURE PASSKEY: Supports FIDO2/WebAuthn, FIDO U2F, Yubico OTP, OATH-TOTP/HOTP, Smart card (PIV), and OpenPGP. That means it’s versatile, working almost anywhere you need it
- PRIMARY & SPARE KEYS: Just like having a spare house key, we recommend buying two YubiKeys - one for daily use and one as a spare. That way you’ll never get locked out of your accounts
That article identifies an important exception: conversations flagged by classifiers may still be used for Anthropic’s internal trust-and-safety models. Incognito therefore reduces one use; it does not promise that the content will never be processed for safety purposes.
Anthropic advises particular caution with Social Security numbers, payment-card and bank details, health records, passwords and other credentials, and confidential business or personal documents. Remove those details before asking Claude to summarize, transform, or analyze material.
Gemini: activity, review, and 72-hour retention
Google’s Gemini Apps Privacy Hub was last updated September 24, 2026. Its notice applies to signed-in Gemini Apps and distinguishes signed-out processing, so first establish which state you are using.
Keep Activity on
With Keep Activity on, chats and shared content are stored in Gemini Apps Activity and used to provide, develop, and improve services, including generative-AI model training. Google says human reviewers may examine some content. Data sent to service providers for review is disconnected from the account and retained for three years, according to the notice.
Gemini activity is auto-deleted after 18 months by default. Google provides choices of three months, 36 months, or no automatic deletion. These choices govern the specified activity store; they do not prove that every derivative, safety record, or separately held copy is gone.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Keep Activity off
Google says future chats do not appear in Gemini Apps Activity and are not used to train models when Keep Activity is off, unless you send feedback. The chats are nevertheless retained for 72 hours to provide responses and protect users and the service. Previously reviewed content is not deleted merely because you turn off account activity.
Turning off Keep Activity does not control retention by other Google services. Some Connected Apps also become unavailable, so check which integrations you rely on before changing the setting.
Rank #4
- POWERFUL SECURITY KEY: The Security Key NFC is the essential physical passkey for protecting your digital life from phishing attacks. It ensures only you can access your accounts.
- WORKS WITH 1000+ ACCOUNTS: Compatible with Google, Microsoft, and Apple. A single Security Key NFC secures 100 of your favorite accounts, including email, password managers, and more.
- FAST & CONVENIENT LOGIN: Plug in your Security Key NFC via USB-A and tap it, or tap it against your phone (NFC) to authenticate. No batteries, no internet connection, and no extra fees required.
- TRUSTED PASSKEY TECHNOLOGY: Uses the latest passkey standards (FIDO2/WebAuthn & FIDO U2F) but does not support One-Time Passwords. For complex needs, check out the YubiKey 5 Series.
- BUILT TO LAST: Made from tough, waterproof, and crush-resistant materials. Manufactured in Sweden and programmed in the USA with the highest security standards.
Temporary chats and broad input coverage
Gemini temporary chats also have a 72-hour retention period. Google’s notice covers more than typed prompts: files, videos, screens, photos, imported chats, connected apps, and some device context can be involved. Recheck permissions before enabling a feature that expands what Gemini can receive.
Microsoft Copilot: separate consumer and work experiences
Microsoft’s consumer Copilot FAQ says conversation activity can be used for model training and personalization. It offers controls for both and says consumer conversation activity is kept for 18 months by default. Microsoft advises against entering confidential or sensitive personal data.
Free tools Windows power users keep installed
One-click scans. No signup required.
Do not merge that policy with Microsoft 365 home apps or work/school Copilot Chat. Microsoft says work or school Copilot Chat provides enterprise data protection when you authenticate with an organizational account: prompts and responses are not used to train foundation models. However, prompts, Bing searches triggered by those prompts, and responses are logged and may be viewed by an IT administrator. Workplace confidentiality therefore depends on your organization’s access rules as well as Microsoft’s training statement.
Are temporary chats really private?
No temporary label should be read as “nothing is stored” or “no person or safety system can process this.” The documented examples show why:
- OpenAI says temporary chats are absent from history and memory and are not used to improve models while temporary.
- Google says Gemini temporary chats are retained for 72 hours.
- Anthropic says classifier-flagged Incognito conversations may still support internal trust-and-safety models.
Use a temporary mode for an appropriate one-off prompt after removing sensitive details. Do not use it as permission to paste passwords, medical records, payment data, or confidential company files.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Review memory, personalization, and connected apps separately
A model-improvement opt-out does not necessarily disable features that save information to personalize later replies. Open the memory or personalization panel, inspect saved items, delete obsolete entries, and disable the feature if retaining those facts is unnecessary.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →Connected calendars, drives, email, messaging, search, and other services can introduce data under their own terms. Remove integrations you do not need and review the scope of each permission. A chat deletion cannot necessarily erase a copy retained by the connected service, and turning off an AI activity setting cannot rewrite that service’s logs.
Best Value
- POWERFUL SECURITY KEY: The YubiKey 5 is a versatile physical passkey that protects your digital life from phishing attacks. It ensures only you can access your accounts.
- WORKS WITH 1000+ ACCOUNTS: Compatible with popular accounts like Google, Microsoft, and Apple. A single YubiKey 5 secures 100+ of your favorite accounts, including email, password managers, and more.
- FAST & CONVENIENT LOGIN: Plug in your YubiKey 5 via USB and tap it to authenticate. No batteries, no internet connection, and no extra fees required.
- MOST SECURE PASSKEY: Supports FIDO2/WebAuthn, FIDO U2F, Yubico OTP, OATH-TOTP/HOTP, Smart card (PIV), and OpenPGP. That means it’s versatile, working almost anywhere you need it.
- BUILT TO LAST: Made from tough, waterproof, and crush-resistant materials. Manufactured in Sweden and programmed in the USA with the highest security standards.
What should I not share with AI?
- Passwords, recovery codes, private keys, API tokens, one-time authentication codes, and security-question answers.
- Bank-account numbers, payment-card numbers, Social Security numbers, tax identifiers, and complete financial statements.
- Health records, medical images, insurance identifiers, therapy notes, and another person’s diagnosis or treatment details.
- Confidential contracts, unreleased financial results, source code, customer lists, legal strategy, credentials, and trade secrets.
- Identity documents, precise home addresses, children’s personal information, and intimate photos or recordings.
- Unredacted workplace files or customer communications when a short, anonymized excerpt would answer the question.
Anthropic specifically lists financial details, health records, passwords or credentials, and confidential documents as material requiring care. Microsoft similarly tells users not to provide confidential or sensitive personal information. Those warnings are practical baseline rules even when another provider’s interface appears more restrictive.
Deletion, retention, and human review: what controls can and cannot do
Deleting a chat is one action, not a universal eraser. A provider may retain information for response delivery, abuse prevention, security, debugging, legal obligations, or a stated review process. Content already reviewed may be disconnected from your account, as Google describes, and may not disappear when you delete activity. A safety classifier may process content even when model-improvement use is disabled, as Anthropic describes.
Before uploading, find the product’s deletion path and determine whether you must separately remove the conversation, file, memory, activity entry, and connected-app permission. If a setting says “future” or “new,” interpret it literally: it does not retroactively change earlier processing.
Use workplace material under workplace policy
Ask your administrator which account and surface you are using, what is logged, who can retrieve logs, how long records are kept, and whether legal or security teams can access them. Do not paste company information into a personal consumer account merely because the task is work-related. Conversely, do not assume that an organizational account makes a prompt invisible to administrators.
Where policy permits AI assistance, provide the minimum excerpt, remove customer identifiers, and record which approved service and account handled the material. If policy does not permit the upload, a training opt-out is not a workaround.
A repeatable pre-prompt privacy check
Keep this six-question check beside your workflow:
- What data is in the prompt or attachment? Identify credentials, financial, health, personal, customer, and confidential business information.
- Who operates the account? Confirm the provider, product surface, region, and whether the account is personal, work/school, or API-managed.
- Which control am I relying on? Verify exactly what the improvement, history, memory, temporary, or connected-app toggle covers.
- What remains possible? Check retention duration, safety processing, human review, feedback exceptions, and administrator access.
- How will I delete it? Know the separate paths for chats, files, memories, activity, and integrations.
- Can I do this with less detail? Prefer redaction, a short excerpt, or a synthetic example over a complete original record.
Privacy is a continuing practice, not a one-time setting
Recheck the privacy notice and controls before enabling uploads, audio or video, screen sharing, imported chats, memory, or a new connected app. Providers change labels, defaults, retention periods, and eligible features. Date-specific statements are useful only when matched to the exact account and product you are using.
The Federal Trade Commission’s Office of Technology wrote in January 2024: “There is no AI exemption from the laws on the books.” The FTC also said that retaining or using consumer data for other purposes without clear, conspicuous notice and affirmative express consent can create legal risk, and warned against hiding material disclosures in hyperlinks, legalese, or fine print. That is general FTC guidance, not individualized legal advice, but it reinforces the practical rule: read the stated purpose and do not assume a privacy control covers more than it says.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutePC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Consumer AI can be useful without receiving your most sensitive records. Minimize first, choose the correct account, inspect each control and exception, and share only what you would be comfortable having processed under those exact terms.
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.
