Free tools Windows power users keep installed
One-click scans. No signup required.
Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
If a conversation must not reach a cloud AI provider, do not paste it into ChatGPT. For less sensitive questions, ChatGPT’s Temporary Chat is the closest built-in alternative to a locked conversation: it stays out of normal history, does not create or update memories, and is not used to improve models while it remains temporary. OpenAI may still keep a copy for up to 30 days for safety, and saving the conversation turns it into a regular chat.
For stronger separation, use a local, protected document or a separately vetted encrypted messenger, then share only a sanitized excerpt if you need AI help. Turn off model improvement when that matches your preference, delete ordinary chats and related files separately, and protect account sign-in with phishing-resistant authentication. Those controls cover different privacy boundaries; none makes already-submitted text disappear.
As an Amazon Associate I earn from qualifying purchases.
Recommended Free Tools
What does “private” mean when you use ChatGPT?
People use “private” to describe several different outcomes. A chat can be hidden from the visible history yet still be retained for safety; excluded from model improvement yet still stored in your account; removed from your account while copies remain in another system; or never sent to a cloud service at all. Decide which boundary matters before choosing a setting.
| Privacy goal | Control that addresses it | What it does not address |
|---|---|---|
| Keep a one-off chat out of normal history | Temporary Chat | A safety copy may remain for up to 30 days |
| Prevent use for model improvement | Data Controls preference | Existing saved chats remain until you delete them |
| Remove a saved conversation from your account | Delete the chat | Separate copies such as uploaded files, shared links, exports, or third-party records |
| Keep text entirely outside a cloud AI service | Local protected workflow | You will not get ChatGPT generation unless you later share sanitized text |
| Stop an intruder signing in | Strong authentication, including a hardware security key | Retention rules for prompts already submitted |
These are complementary controls, not interchangeable ones. A setting that changes training or model-improvement use does not erase a chat, and a login key cannot retract data that has already left your device.
#1 Best Overall
- 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
Is there a locked-chat mode in ChatGPT?
ChatGPT does not provide a separate, password-locked conversation container in the privacy controls described here. If you need a one-off conversation with reduced persistence, use Temporary Chat. If you need text to remain inaccessible to the service itself, keep the material in a local workflow instead.
What Temporary Chat changes
OpenAI describes Temporary Chats this way: “Temporary Chats won’t appear in your chat history, won’t create memories, and won’t be used to train our models.” In practical terms, the conversation is not listed with your normal chats, and it does not add new memories while it is temporary. It is also excluded from model-improvement use while temporary.
That is a retention reduction, not a promise that no copy exists. OpenAI says it may keep a copy of a temporary chat for up to 30 days for safety purposes. The conversation is still submitted to OpenAI’s systems so the model can answer, and any connected service involved in an action can have its own retention policy.
What Temporary Chat may still use
Depending on the personalization choice you make, a temporary conversation can use existing memories, custom instructions, or plugins. The exact options vary by app and account version. Before sending sensitive text, choose the least personalized option available and check whether any connected tools or actions are enabled.
Why saving a Temporary Chat changes the boundary
Saving a Temporary Chat converts it into a regular chat. Once saved, it is subject to ordinary account settings and appears in your normal conversation management. Treat the save control as a deliberate change in retention, not as a harmless way to bookmark a temporary answer.
How do I start a Temporary Chat safely?
Use the Temporary Chat control before you type or paste anything. Labels and locations can differ between the web app, desktop app, and mobile app, so confirm that the temporary indicator is visible before sending the first message.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minute- Open a new conversation. Use the new-chat control, then open the model or conversation options where your app lists Temporary Chat.
- Select Temporary Chat. Verify the temporary label or banner is shown. If the app offers personalization choices, pick the least personalized setting that still meets your needs.
- Check connected features. Turn off plugins, actions, or integrations you do not need. A third-party action can have separate storage and deletion rules.
- Minimize the data. Replace names, account numbers, credentials, exact addresses, health identifiers, and proprietary passages with neutral placeholders before sending.
- Ask only for the transformation you need. A short, sanitized excerpt creates less exposure than an entire file or complete case history.
- Do not save the result. Saving changes the conversation to a regular chat with normal retention behavior. Copy only the non-sensitive answer you need into your own protected notes.
Temporary Chat is most useful for one-off assistance such as explaining a generic error message, rewriting text after personal details have been removed, or brainstorming from a de-identified description. It is not the right destination for credentials, private keys, unredacted medical records, confidential legal files, or trade secrets that a cloud provider must never receive.
How do I stop ChatGPT from using my chats for model improvement?
Open the account settings, go to Data Controls, and turn off the model-improvement option if that matches your risk tolerance. The wording and menu placement can change as the interface is updated, but the relevant control is the setting that governs whether your conversations may be used to improve models.
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
OpenAI’s Data Controls guidance is explicit that turning off model improvement “does not delete or hide saved chats.” Existing conversations remain in your account until you remove them through chat management. This setting is therefore an ongoing preference, not a cleanup command.
When this setting is useful
- You regularly use ordinary chats and want a standing preference against model-improvement use.
- You understand that the conversations still exist in your account and may be subject to the service’s operational retention and safety processes.
- You want an additional safeguard alongside redaction and selective use of Temporary Chat.
When it is not enough
- You need the text to stay entirely off a cloud service.
- You need to remove old conversations, files, or links already stored in the account.
- You are trying to protect the account from an attacker; that requires stronger authentication and device security.
How do I delete a private conversation and its related copies?
Delete ordinary chats from the conversation list or chat-management screen, then verify that the item no longer appears. Deleting the visible chat is only one part of a complete cleanup because other objects can have separate controls.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
- Delete the conversation. Use the chat’s menu and choose the delete action, then confirm.
- Review uploaded files. Check the file area or Library and remove documents that contain the same sensitive material. A deleted chat does not automatically prove that every uploaded item has been removed.
- Revoke shared links. If you created a link for another person, open shared-link management and disable it. Anyone who already copied the content may retain their own copy.
- Check connected apps and actions. Review services that received data through an integration. Their retention and deletion procedures are separate from ChatGPT’s chat list.
- Remove local exports. Delete downloaded transcripts, screenshots, browser downloads, and backups from devices and cloud drives you control.
Deletion and model-improvement preferences solve different problems. Set the preference for future use, then perform the deletion work for material that is already saved.
What should I redact before sending a conversation?
Redaction reduces the harm if a prompt is retained, exposed through an account compromise, or copied into a support workflow. Remove direct identifiers and anything that allows a person or system to be reconstructed from the text.
High-priority removals
- Passwords, API keys, recovery codes, private certificates, and session tokens.
- Bank, payment, tax, insurance, and government-identification numbers.
- Full names combined with dates of birth, exact addresses, phone numbers, or email addresses.
- Detailed health information, medical record numbers, and appointment identifiers.
- Unpublished source code, customer lists, contracts, deal terms, and proprietary research.
- Exact location data, travel schedules, access instructions, and photographs containing identifying metadata.
Safer substitutions
- Use labels such as
CLIENT_A,DATE_1, andACCOUNT_ID_REDACTED. - Round dates or amounts when precision is not required for the answer.
- Paste the smallest excerpt that demonstrates the problem rather than an entire document.
- Describe a credential format without revealing a live credential.
- For code, replace hostnames, tokens, package-internal names, and customer data while preserving the error and relevant logic.
Redaction is not encryption. If disclosure itself is unacceptable, do not send the material and rely on a local protected workflow.
When should I keep the conversation entirely outside ChatGPT?
Use a local protected workflow when a cloud provider must not receive the text, even temporarily. This is the only option in the comparison that keeps the original material out of ChatGPT by design.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →A practical local workflow
- Create a protected working file. Use full-disk encryption and an account with a strong device passcode. Keep the file in a location covered by your normal backup policy.
- Limit access. Close sharing links, remove unnecessary collaborators, and avoid syncing the file to unmanaged devices.
- Work offline when feasible. Disconnect network access while drafting or reviewing the most sensitive portions if your tools support that safely.
- Sanitize before optional AI assistance. Make a separate copy containing only the facts needed for the question, with identifiers and secrets removed.
- Track the copies. Know where the original, sanitized copy, exports, and backups live, and delete temporary copies when the task is complete.
Local storage shifts responsibility to your device, operating system, backup provider, and anyone with physical or administrative access. Use encryption, timely updates, malware protection, and a recovery plan; “local” is not automatically secure.
When an encrypted messenger is a better fit
If the goal is to discuss sensitive material with another person rather than generate text, choose a separately vetted encrypted messenger whose threat model and retention practices you understand. Verify who controls the service, how keys are managed, whether backups are encrypted, and what metadata remains. An encrypted messenger can protect a conversation in transit and at rest according to its design, but it does not provide ChatGPT’s generation features.
Rank #3
- NIST Certification: FIPS 140-3 validated for government and regulated organizations (Overall Level 2, Physical Security Level 3).
- Works with 1000+ Accounts: Supported by Google and Microsoft accounts, Identity Access Managers, password managers and 1000+ popular services. It works with operating systems and browsers including Windows, macOS, Chrome OS, Linux, Chrome, and Edge.
- Fast & convenient login: Plug in your YubiKey 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.
- 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. Made in Sweden with the highest security standards.
Can a hardware security key make ChatGPT conversations private?
A hardware security key protects account access; it does not change what happens to prompts after submission. Yubico documents that the YubiKey 5 NFC supports FIDO2/WebAuthn, described as a hardware-bound passkey, and NFC tap authentication. Its compatibility list includes OpenAI, so it can be a phishing-resistant sign-in option where your account and device support it.
What the key protects
- It makes a stolen password less useful because the attacker also needs the physical key and the authentication ceremony.
- It reduces exposure to phishing pages that can capture passwords and one-time codes.
- It can protect the account that contains saved chats, files, and settings.
What it cannot protect
- It cannot erase a prompt already sent to ChatGPT.
- It cannot shorten the safety-retention window for Temporary Chat.
- It cannot override Data Controls, delete chats, or control a third party that received data through an action.
Keep a recovery method and, where appropriate, a spare key stored securely. Strong login protection is valuable, but it belongs alongside data minimization and retention controls rather than replacing them.
Which privacy option should I choose?
Match the option to the consequence of disclosure, not to the convenience of the interface.
| Situation | Recommended approach | Reason |
|---|---|---|
| Generic question with no identifying details | Normal chat or Temporary Chat | The content carries little personal or confidential information |
| Personal question where a short-lived transcript is acceptable | Temporary Chat, least personalized setting, and redaction | Keeps the chat out of normal history and model improvement while temporary |
| Ongoing use where you do not want chats used for model improvement | Data Controls preference plus regular deletion reviews | Changes the future-use preference without pretending saved chats are gone |
| Confidential business, legal, medical, or security material | Local protected workflow; share only sanitized text if necessary | The original never needs to reach ChatGPT |
| Concern about account takeover | Hardware security key or another strong phishing-resistant method | Protects access to the account, not prompt retention |
What are the common privacy mistakes and fixes?
Mistake: treating hidden history as deletion
Fix: Remember that Temporary Chat can have a safety copy for up to 30 days. Do not use it for material that may not be disclosed to the provider at all.
Mistake: turning off model improvement and assuming old chats vanished
Fix: Delete existing chats and separately review files, links, integrations, and exports.
Mistake: saving a temporary conversation for convenience
Fix: Copy only a sanitized answer into a protected note. Saving converts the temporary chat to a regular one.
Mistake: pasting a complete document to get help with one paragraph
Fix: Extract the smallest relevant passage and replace identifiers, secrets, and proprietary details.
Mistake: buying a security key to solve a retention problem
Fix: Use the key for account takeover resistance, then handle retention with Temporary Chat, Data Controls, deletion, or a local workflow.
Rank #4
- 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.
A simple policy for private conversations
Adopt a short rule that everyone using the account can follow: classify the material first. If disclosure to a cloud provider is unacceptable, keep it local. If a sanitized, one-off question is acceptable, use Temporary Chat, select the least personalized option, avoid connected actions, and do not save the result. For routine use, set Data Controls according to your preference, schedule deletion reviews, and protect sign-in with strong authentication.
This policy keeps the boundaries distinct: Temporary Chat reduces persistence, Data Controls changes model-improvement preference, deletion removes account items you select, local tools keep originals outside ChatGPT, and a security key protects the door to the account. Choosing the right boundary before sending is more reliable than trying to recover privacy afterward.
Frequently Asked Questions
Can Temporary Chat read memories or custom instructions I already created?
It can, depending on the personalization choice active when you start the conversation. Review those options and disable unnecessary personalization before sending sensitive material; Temporary Chat’s no-memory promise refers to creating or updating memories from that chat.
What happens if I accidentally save a Temporary Chat?
Saving converts it into a regular conversation subject to normal account settings. Delete the chat from your history, then check related files, shared links, exports, and connected services for additional copies.
Is an encrypted messenger always safer than ChatGPT?
Not automatically. Safety depends on its encryption design, key handling, backups, metadata, device security, and the people who can access the endpoints. Compare those properties with your threat model and use a local workflow when the text must never reach a cloud service.
Will deleting a chat remove a file I uploaded in it?
Do not assume so. Review the service’s file or Library area and remove the uploaded item separately, then revoke any shared links and delete local downloads or backups you created.
Do I need a security key if I never save chats?
A key can still prevent an attacker from entering the account and viewing settings or other stored data. It does not alter retention for prompts you submit, so use it as account protection alongside careful data handling.
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.
