Valentius Kryptix

No — OpenAI does not sell or share your ChatGPT conversations with advertisers. Ads can be shown based on the current conversation’s context, and — only if you’ve turned on ad personalisation — a limited set of broader signals from your ChatGPT history. Either way, advertisers themselves never receive the underlying chats, your name, or other personal details; they get aggregate reporting like impression and click counts, nothing more granular. That’s the platform mechanic. There’s a separate, genuinely open question about how India’s data protection law applies to any of this, which the second half of this article covers honestly rather than confidently.

Key takeaways:

• Advertisers never see your actual conversations, chat history, memories, or personal details — under any settings.
• Your current conversation can always factor into ad relevance. Broader signals (past chats, memory) only apply if you’ve explicitly turned on ad personalisation.
• Ads don’t run near personal health, mental health, or political conversations, regardless of your settings.
• You can turn ad personalisation off entirely from within ChatGPT’s settings.
• India’s DPDP Act consent requirements as applied to ad personalisation specifically remain a genuinely unsettled area — this article states what’s known and flags what isn’t, rather than pretending certainty either way.

What Data Is Used to Choose Your Ads

None of this applies if you’re not seeing ads in the first place. Ad personalisation, and everything below, is only relevant to Free and Go tier accounts — Plus, Pro, Business, Enterprise, and Edu subscribers don’t see ads at all, so there’s no ad-related data use to worry about on those plans regardless of any setting. If you’re on a paid tier wondering whether this article applies to you, it doesn’t.

Two different signals can feed into which ad you see on an ad-eligible account, and they’re gated very differently.

Current Chat Context (Always)

Whatever you’re currently discussing can factor into ad relevance regardless of your settings — that’s the baseline mechanic the whole ad product runs on. Ask ChatGPT to help you compare accounting software and a relevant ad may appear below the answer; the system used the conversation itself to judge that relevance, the same way it would use a search query on any other platform. This applies every time, with no opt-out, because it’s the core targeting mechanism, not an optional enhancement.

Broader Personalisation Signals (Only If Enabled)

Beyond the current conversation, OpenAI’s documentation indicates that — only if you’ve turned on ads personalisation — a limited set of broader signals from your wider ChatGPT experience can also apply, including things like past chats and memory. With personalisation off, none of that broader history factors in; relevance comes from the current conversation alone. This is opt-in, not opt-out, and it’s worth checking your own setting rather than assuming either way — see the controls section below.

What Advertisers Never See

Regardless of your personalisation setting, there’s a hard boundary between what the system can use internally to judge relevance and what an advertiser actually receives. OpenAI’s Help Center is explicit that advertisers don’t get your chats, your chat history, your memories, or other personal details — full stop.

An advertiser can learn “this campaign got 10,000 impressions and 300 clicks.” They cannot see the 10,000 conversations that produced those impressions, and they can’t identify you individually within that number. Worth being precise about the boundary this table draws, since “not selling your conversations” and “processing nothing at all” aren’t the same claim — the system still has to process conversation context internally to decide an ad is relevant in the first place. What it doesn’t do is hand that processing over to the advertiser as a byproduct.

For the fuller mechanics of how ad relevance is actually calculated — context hints, the auction, personalisation signals — see our guide to how ChatGPT Ads work.

How to Turn Off Ad Personalisation

Ad personalisation is a setting you control, not something permanently on. Inside ChatGPT, look under Settings for a data controls or ads personalisation option — the exact menu path has moved before as OpenAI updates the app, so if what’s described here doesn’t match what you see, search “ads” or “personalisation” within Settings directly rather than assuming the feature has disappeared. Turning it off doesn’t turn off ads entirely on the Free or Go tier — it means ad relevance falls back to your current conversation only, with none of the broader signals in play.

Ads Are Never Shown Near These Topics

Regardless of any setting, OpenAI states that ads aren’t eligible to appear near personal health, mental health, or political conversations. This is a category-level restriction, not something a user has to opt into for protection — it applies by default. Regulated categories like financial services and healthcare advertising can still exist on the platform, but only through additional review on the advertiser’s side; the restriction is about where ads can’t appear near sensitive personal topics, not a blanket ban on entire industries advertising at all.

Worth being precise about what this restriction actually covers: it’s about the immediate conversational context an ad might appear next to, not a permanent flag on your account because you once discussed a sensitive topic. Ask ChatGPT something health-related in one conversation and something unrelated in the next, and the second conversation is judged on its own terms — the restriction is contextual, applied per-conversation, not a lasting label attached to you as a user.

One nuance worth flagging for advertisers specifically, not just users: a category being eligible to advertise at all is a different question from being allowed to use sensitive personal information for targeting within it. A finance or healthcare brand clearing the category-level review isn’t automatically cleared to target based on someone’s disclosed financial stress or health condition — those are two separate gates, not one. If you’re evaluating this specifically for a healthcare or finance brand, our guide to restricted category rules covers what that additional review actually involves.

How Long Is Ad Data Kept?

Worth not treating this as one database with one answer. Your regular chat history, your ad personalisation settings, and ad interaction/measurement data are handled as distinct data categories, not a single blob with a uniform retention period — so “how long is my data kept” doesn’t have one universal number across all of it. For ad interaction data specifically, OpenAI’s documentation points to a retention window in the region of 30 days. Treat that as the current stated figure for that one category rather than a blanket answer for everything above — retention periods are the kind of detail platforms revise, and different data types genuinely follow different rules, so check OpenAI’s current Help Center directly if the exact figure matters for a compliance decision you’re making.

The Open Questions — India’s Regulatory Gap

Everything above describes what OpenAI states its system does. It’s worth being equally direct about what’s genuinely unresolved, rather than presenting platform policy as the whole picture. A caveat before this section: none of what follows is legal advice. If ad personalisation consent is a real compliance question for your business, talk to an actual lawyer, not a marketing blog.

Consent Standards Under India’s DPDP Act

India’s Digital Personal Data Protection Act sets requirements around consent, purpose limitation, and how personal data can be used — but the detailed rules governing exactly how those requirements apply to behavioural ad personalisation specifically are still being worked out in practice, not fully settled the way, say, GDPR’s ad-consent framework is in the EU. Whether ChatGPT’s opt-in personalisation toggle meets the notice-and-consent bar the DPDP Act ultimately requires is a genuinely open question, not something this article — or any marketing content — can responsibly answer with confidence. If you’re a business in a consent-sensitive category, that uncertainty is itself the useful fact: don’t treat a platform’s own opt-in toggle as automatic proof of regulatory compliance on your end.

The “Vulnerability Inference” Concern

Independent commentary on the India rollout — Medianama’s coverage in particular — has raised a fair concern worth stating plainly rather than dismissing: even with health, mental health, and political topics excluded from ad placement, a user’s broader conversational patterns across a session could still reveal vulnerability-adjacent context that factors into more ordinary-looking ad matching. Someone mentioning financial stress while asking about budgeting apps, for instance, sits in a greyer zone than an explicit mental-health conversation, and it’s not obviously covered by a category restriction built around named sensitive topics. OpenAI’s stated boundaries address the clearest cases; whether they fully address subtler ones is a fair question critics have raised, and this article isn’t going to pretend the answer is a confident yes just because reassurance would read better.

Worth noting for fairness: this isn’t a criticism unique to OpenAI. Any ad-supported product that infers relevance from user behaviour faces some version of this same edge case, and OpenAI hasn’t specifically responded to the vulnerability-inference framing as far as public reporting shows. That doesn’t resolve the concern, but it’s a reason to treat it as an industry-wide open question rather than evidence of bad faith specific to ChatGPT Ads.

How to Check and Control Your Own Ad Settings

Practically, three things are worth doing if you want to know exactly what you’re opted into: check your current ads personalisation setting inside ChatGPT’s Settings menu rather than assuming a default, review whichever broader data controls govern chat history and memory retention since personalisation draws on those same settings, and revisit both periodically, since OpenAI has changed settings menus before and is likely to again as the ads product matures. None of this requires technical skill — it’s a five-minute check, and it’s the most direct way to know your own answer rather than relying on a general article like this one.

What This Means If You’re Evaluating ChatGPT Ads as a Brand

Everything above is written for a user protecting their own privacy. If you’re on the other side — a business considering ChatGPT Ads, or a compliance team asked to sign off on it — the useful question isn’t “is ChatGPT Ads privacy-safe,” which is really a question about OpenAI’s side of the platform. It’s “is our complete customer journey through this channel appropriate for the data we collect and the customers we serve,” which includes the parts that have nothing to do with OpenAI at all.

That distinction matters because the moment someone clicks your ad and lands on your website, an entirely separate data relationship starts — your own analytics, your CRM, your remarketing pixel, whatever conversion tracking you’ve set up. None of OpenAI’s privacy commitments about ChatGPT conversations extend to what happens on your landing page. A compliance review that stops at “does OpenAI share conversations with us” and never looks at what your own site collects after the click has only checked half the picture.

A short list worth actually asking, rather than assuming: what does your landing page collect from a visitor, and does it match what your privacy notice says it collects? If the campaign touches a sensitive category, does the landing page itself avoid capturing health or financial details it doesn’t need? Are your own conversion-tracking and remarketing tools configured in a way your compliance team has actually reviewed, or inherited from a template? None of these are ChatGPT Ads-specific questions — they’re the same questions a careful team should ask before running Google or Meta campaigns too — but it’s easy to assume a new platform’s privacy documentation covers your own website’s behaviour when it doesn’t.

Evaluating ChatGPT Ads for a healthcare, finance, or otherwise regulated brand, and need the privacy and compliance picture straight before you commit budget? Valentius Kryptix can walk you through what a compliant campaign setup actually looks like. Talk to our team →

Frequenty Asked Questions