OpenAI Changed Its Privacy Policy 4 Times in One Week. Here Is What Actually Changed.

The June Privacy Rewrite

Between June 6 and June 12, 2026, OpenAI revised its privacy policy four times. Not once. Four times in seven days. Each revision was incremental. Together, they represent a significant shift in how OpenAI handles advertising data, user controls, and content monitoring.

Most users saw none of it. Privacy policy changes do not trigger push notifications. They do not appear in the ChatGPT interface. Unless you were comparing document versions side by side, the changes were invisible.

Here is what actually changed, in order.

June 9: Ad Personalization Controls Disappear

OpenAI removed explicit descriptions of ad personalization controls available to Free and Go users. Before this change, the privacy policy described specific settings users could access to manage how their data was used for ad targeting. That language was removed.

The authorization to personalize ads remained. The documented controls disappeared. If you are a Free or Go user, OpenAI can still personalize the ads you see based on your chat content and ad interactions. You just no longer have a documented path to manage it.

This matters because OpenAI rolled out advertising to Free and Go users earlier in 2026 with a clear promise: ads would be personalized, but users would have controls. Removing the documentation of those controls without explanation changes the relationship.

June 12: Advertiser Partnerships and Contact Scanning

Three days later, OpenAI made a larger revision. Advertiser data partnership language was removed from the privacy policy. Previously, the policy disclosed that OpenAI may receive information from advertisers about purchases users make, used to measure ad effectiveness. That disclosure was pulled.

In the same revision, OpenAI added language about contact scanning and content monitoring. The policy now describes capabilities for scanning user contacts and monitoring content that were not documented in prior versions.

Read that again: in a single revision, OpenAI removed transparency about advertiser relationships and added new data collection capabilities. The net effect is more data collection with less disclosure.

The Pattern Is Incremental

No single change is dramatic enough to make headlines. That is what makes it worth tracking.

In February 2026, OpenAI updated its privacy policy to formally authorize ad personalization. Ads would appear for Free and Go users. Paid users would not see ads. Users would have controls.

In March, the policy added that OpenAI may receive purchase data from advertisers. A new data flow, disclosed transparently.

In April, the policy was revised again with expanded language about data usage for ad measurement.

By June, the controls were being removed and the advertiser disclosure was pulled. Each step was small. The trajectory is clear: broader data use, less documentation of user agency, and expanding collection capabilities.

This is governance drift. Not a single dramatic change, but a series of incremental revisions that cumulatively reshape the terms under which 800 million weekly users operate.

What OpenAI Actually Governs

OpenAI maintains 15 separate governance documents. Most users have read zero of them. Here is what they cover:

Your conversations. The privacy policy governs how OpenAI collects, stores, and uses the content of your ChatGPT conversations. For Free and Go users, conversation content can be used to personalize advertising. For paid users, different terms apply.

What you can build. The usage policies define what developers can and cannot do with OpenAI APIs. These 60 provisions govern every application built on GPT, DALL-E, Whisper, and Sora. A change to usage policies is operationally a change to thousands of applications.

Your legal rights. OpenAI requires mandatory arbitration with class action waivers. 15 provisions across the terms require individual dispute resolution. If OpenAI changes its data practices in ways that affect millions of users, each user must pursue their claim individually.

Enterprise data handling. Separate Business Terms and a Data Processing Addendum govern enterprise customers. These include different data retention, processing, and privacy commitments than consumer terms. Enterprise customers operate under a different governance framework than Free and Go users.

Three Things Free and Go Users Should Know

Your conversations inform ad targeting. This has been the case since February 2026. OpenAI personalizes ads using signals from your chat content and prior ad interactions. Advertisers do not see your conversations directly, but OpenAI uses conversation context to decide which ads to show you.

The controls you were promised may not be documented anymore. The June 9 and June 12 revisions removed descriptions of ad personalization controls from the privacy policy. Whether the controls still exist in the product interface is a separate question from whether the policy documents them. Check your ChatGPT settings directly.

Paying $20/month changes your privacy terms. Plus, Pro, Business, Enterprise, and Edu users do not see ads and operate under different data handling terms. The advertising and ad personalization provisions apply specifically to Free and Go tiers. Upgrading is not just a feature decision. It is a privacy decision.

What Developers Should Know

API terms change independently from consumer terms. OpenAI maintains separate Service Terms, Business Terms, and API Data Usage Policies. Changes to the consumer privacy policy do not necessarily affect API customers, but usage policy changes affect everyone.

Usage policy changes are operationally your problem. When OpenAI adds a content restriction or modifies what applications can do with its models, every dependent application must comply. OpenAI does not notify individual developers of usage policy changes.

Your dispute resolution options are limited. Mandatory arbitration applies to API customers under the Service Terms and Business Terms. If OpenAI restricts your API access or changes rate limits, your recourse is individual arbitration under terms OpenAI drafted.

The Full Timeline

View the complete OpenAI governance timeline

Why This Matters Beyond OpenAI

OpenAI is not the only AI platform changing its terms. But it changes them more frequently than any other AI provider. Anthropic has generated 3 governance events in the same period. Google Gemini has generated 19. OpenAI has generated 36.

The frequency matters because each change is easy to miss individually. Nobody reads a privacy policy revision email. But when you see the full timeline, the cumulative trajectory becomes clear: OpenAI is actively reshaping its governance framework around advertising, data collection, and user controls. Whether that trajectory aligns with what users expect depends on whether anyone is paying attention.

Primary Sources

OpenAI governance profile (15 documents, 360 provisions)

OpenAI arbitration provisions

OpenAI privacy provisions