Can locked ChatGPT chats hide your job search?

What do locked ChatGPT chats actually hide?

Locked chats hide a conversation from whoever picks up your unlocked phone — and from nobody else. OpenAI announced the feature on 20 October 2026, and the mechanics are simple: open a chat, tap the three-dot menu, and the thread moves into a dedicated Locked Chats section. It drops out of your regular history, your pinned chats and your search results, and the title stays hidden until you unlock it. On mobile you authenticate with Face ID, Touch ID or an Android fingerprint sensor; on web and desktop you type a four-digit PIN. It's rolling out gradually across Free, Go, Plus, Pro, Business, Enterprise and Edu plans, though not ChatGPT for Healthcare.

What locking doesn't do matters far more for a job search. It isn't end-to-end encryption and it isn't a separate encrypted vault. It doesn't shorten how long your data is kept, doesn't change the "Improve the model for everyone" setting, and doesn't erase anything Memory already absorbed before you flipped the lock on. Locked Chats does get its own pool of memory, so a salary figure you mention after locking won't resurface in an unrelated thread about dinner recipes. You also can't lock everything: temporary, archived, group and shared-link chats are excluded, as are chats with scheduled tasks or meeting recordings. And the lock is per conversation, not app-wide — whatever you forget to lock sits in plain view.

One trap is worth knowing before you move your whole search behind a PIN: resetting that PIN permanently deletes every locked chat and file protected by the old one. No support ticket brings them back. So if a locked thread holds the only copy of your tailored CV, your interview notes and that painstaking list of target companies, you're one forgotten PIN away from starting over. Keep the files on your own machine and treat the locked chat as scratch space. My honest read: this is a privacy screen against shoulder-surfing, not a security control. Genuinely useful on a shared iPad at home, close to worthless against the parties job seekers actually worry about.

Can your employer see your ChatGPT job-search chats?

No, locked chats hide nothing from your employer. The lock sits inside ChatGPT's interface on the account you're signed into; it does nothing at the account-administration layer and nothing at the device or network layer. If you're signed into a work ChatGPT account, your conversations live in a workspace your company controls. On Enterprise and Edu plans, workspace admins can retrieve conversations, uploaded files, memories, GPT configurations and user records through the Compliance API, which exists so regulated firms can feed ChatGPT activity into their eDiscovery, DLP and SIEM tooling. Access to message content has to be granted by a workspace owner, and the compliance logs are append-only — deleting a chat doesn't scrub the record.

On the Business plan the picture is genuinely murkier, and I'd rather say so than pretend otherwise. OpenAI's own documentation pulls in two directions: the Business FAQ says admins and owners can't see members' private chats by default, while the enterprise privacy page describes workspace owners being able to view, access, export and delete end-user conversations. In practice that access gets exercised during litigation, an investigation or an audit rather than on a Tuesday afternoon — your team lead is not browsing your threads for sport. But capability is what you plan around, not intent. If the account was provisioned by your company, assume somebody could read it and move the conversation elsewhere.

The device matters just as much as the account. A company laptop usually ships with monitoring software already installed, and a managed Chrome or Edge profile gives IT visibility into browsing activity with nothing visible in the taskbar. Type chrome://management or edge://management to see whether a policy is applied. Corporate DNS and web filtering log the domains you reach with timestamps whether or not you're in incognito — private browsing only stops your own browser saving history locally. Connect a personal phone to office Wi-Fi, or leave the work VPN running at home, and you hand over the same visibility. Locking a chat changes none of this, because the lock never leaves ChatGPT.

Is it safe to paste your CV into ChatGPT?

Pasting your CV into ChatGPT is low-risk on a personal account with training switched off — the dangerous material is everything you type around it. A CV is a document you already hand to strangers. Your current salary tied to your employer's name, the real reason you're leaving, a candid read on your manager, your shortlist of target companies, the internal metrics you're quoting as achievements: none of that appears on the CV, and all of it tends to end up in the thread. That's the conversation you don't want appearing in a screen share, a workspace export, or on a colleague's monitor when you pull up ChatGPT to show them something else entirely.

Know the defaults, because they differ by plan. On Free, Plus and Pro, conversations may be used to train models unless you opt out in settings; Business, Enterprise and Edu don't train on your content by default. Delete a chat and it disappears from your view immediately and is scheduled for permanent deletion within 30 days, with carve-outs for legal and security obligations and for data already de-identified. Temporary Chats stay out of history and create no memories, but a copy may be retained up to 30 days for safety review. The sloppiest exposure is also the easiest to fix: a share link stays readable by anyone holding it until you revoke it.

So redact by habit. Strip your home address, date of birth, national ID number and referees' phone numbers before you paste — none of which belongs on a modern CV anyway. Keep NDA material out: client names, unreleased products, internal revenue or headcount figures. Describe your employer as "a mid-size fintech" rather than naming it when you're explaining why you want out. Say "targeting roughly 25% above my current band" instead of printing both numbers in one sentence. Be stricter still if you've connected an AI agent to your inbox or job-board logins — the permission checklist for agents handling your CV is worth ten minutes before you grant anything access.

How do you set up a confidential job search in an AI assistant?

A confidential AI job search comes down to four decisions: personal account, personal device, personal network, and one locked thread for the work. Sign out of the work ChatGPT account entirely rather than switching tabs, and run your sessions on your own phone or laptop with the corporate VPN disconnected. A separate browser profile that isn't signed into any work identity keeps the two worlds from bleeding together. If a company machine is all you have, that's a hard constraint and no setting fixes it — use your phone on cellular data instead. Then lock the thread holding your CV drafts and target list, so a borrowed phone never becomes an accidental resignation announcement.

Next, clean up what's already stored. Locking a chat today doesn't remove what Memory learned from it yesterday, so open your saved memories and delete every entry naming your employer, your salary or your intention to leave. Turn off "Improve the model for everyone" if you'd rather your drafts stayed out of training. Audit your shared links and revoke anything you created months ago and forgot. Request a data export once if you want to see exactly what's sitting in the account. Then use Temporary Chats for one-off pastes you don't need to keep — a job ad you're sizing up, a quick bullet rewrite — and reserve the locked thread for work you'll come back to.

Finally, close the small gaps that leak more than people expect. Switch off message previews in your lock-screen notifications so a chat titled "resignation letter draft" can't surface while your phone lies face-up in a meeting. Don't leave files named CV_for_Competitor_final.pdf in a Downloads folder you screen-share from. And if the privacy question really bites — you're mid-negotiation, or your industry is small enough that one leak ends the search — what on-device AI actually keeps private is worth a read, because a local model genuinely keeps your address and salary off anyone's servers, even if it can't tell you how a parser will read your file.

Where should the CV work actually happen?

Use the assistant for thinking and drafting, and a dedicated CV analyser for the verdict on your real file — because a chat window reads the text you pasted, not the PDF a parser receives. That gap is where confident-sounding advice goes wrong: your two-column layout, your icons, your text boxes and the contact details tucked into a header never made it into the prompt, so the model can't flag any of them. HRLens's free CV analysis takes the actual file — PDF, Word or even a photo, with Hebrew OCR — and scores it out of 100 with five category scores covering experience, tech stack, impact and ownership, clarity and structure, and ATS compatibility, plus a visual layout analysis.

It also narrows what you have to disclose anywhere. You upload one file and paste the text of the job ad — the text, not the URL, since LinkedIn and Indeed block fetching — and the analysis lists the skills you're missing and generates ATS-friendly rewrites in six templates that mirror the ad's keywords. Your salary expectations, your reasons for leaving and your target-company shortlist never have to enter the conversation at all. The free tier gives you the score, the category breakdown, your strengths and the layout read; the $3.99 full analysis adds the fix list, why you might be rejected, your blind spots, hiring probability and a salary estimate.

If the CV needs rebuilding rather than patching, the chat-based CV builder takes an old file or a plain description of your experience, refines it by conversation, and exports to PDF or Word in six templates — Hebrew CVs written in Hebrew and exported right-to-left. The split I'd recommend is clean: brainstorm and phrase in your AI assistant behind a locked chat, keep identifying details and workplace gossip out of it, then let a tool that reads the finished file tell you whether automated screening will. Locked chats protect you from the person holding your phone. Nothing protects you from a CV the parser can't read.

Frequently asked questions