BACK TO HOME
openai

openai

General Serviceopenai.com

Privacy Conclusion

"OpenAI collects extensive user data including prompts, location, and device identifiers for model training and ad targeting. Broad sharing with business account administrators, marketing partners, and government authorities, combined with indefinite retention for banned accounts and vague policy change procedures, creates elevated privacy risk."

Risk Score
62
ELEVATED

Take Action

7 Privacy Improvements Available

View Actions

openai Privacy Concerns & Scorecard

Privacy Risk Analysis
Data CollectionCRITICAL

OpenAI collects and uses user prompts, uploaded files, images, audio, and video content to train AI models. This represents collection of user-generated content at scale without default opt-in consent.

Third-Party SharingCRITICAL

User content and data shared broadly with vendors, service providers, marketing partners, affiliates, and government authorities. Business account administrators receive unrestricted access to all user content and account controls.

Tracking & AdsHIGH

Free and Go users subject to personalized advertising based on collected data. Marketing partners receive data through cookies and similar technologies; limited opt-out mechanism available but not prominently surfaced.

Retention & ControlHIGH

Data retention periods are vague and indefinite for certain categories. Banned accounts retain data indefinitely; legal obligation retention has no specified time limit.

Transparency & RightsHIGH

Policy changes can be made unilaterally by OpenAI with only publication of the updated version as notice. Users have no explicit right to opt-in to new practices or receive advance notice.

Recommended Actions

Ad data clear

  1. 1Open ChatGPT on web or iOS
  2. 2Go to Settings and select Ad Controls (or visit chatgpt.com/#settings/Advertising)
  3. 3Select the option to clear data used for ads
Access Settings

Ai training opt out

  1. 1Click your profile icon in ChatGPT
  2. 2Select Settings
  3. 3Go to Data Controls
  4. 4Turn off the toggle for "Improve the model for everyone"
Access Settings

Temporary chat privacy

  1. 1ChatGPT offers a Temporary Chat mode where conversations are not saved to history, are not used to train AI models, and are deleted from OpenAI's systems after 30 days. This provides a privacy-preserving way to interact with ChatGPT without contributing to model training.
Access Settings

Automate your privacy

Connect your accounts to TrueTerms to automate these privacy actions and monitor policy changes.

Data Collection & Tracking

Personal Information Collected

Identity Data

Account creation, authentication, payment processing, and identity verification for privacy rights requests

Location Data

Service personalization, compliance, and usage analytics

Device & Technical Data

Service delivery, fraud prevention, security, and device-level personalization

Network & Cookie Data

Session management, analytics, advertising personalization, and cross-site behavior tracking

Usage & Behavior Data

Service improvement, AI model training, analytics, and personalization

Contact & Social Data

Invite features, social integration, and cross-referencing against user database

Content & Communications

Service delivery, personalization, model training, and content moderation

Transaction & Payment Data

Payment processing, billing, fraud detection, and financial reporting

Cross-Platform Tracking

OpenAI uses cookies and device identifiers to track users across third-party websites for targeted advertising (Free/Go users only). The Atlas browser feature collects browsing history. Device contact uploads are cross-referenced against OpenAI's entire user base to identify and potentially target non-users for future sign-up campaigns.

Tracking Methods:

First-party Cookies

Cookies set by OpenAI.com to maintain sessions and store user preferences

Third-party Cookies & Pixels

Tracking pixels and cookies from analytics providers (e.g., Google Analytics) and ad networks placed on OpenAI properties and third-party sites to track user behavior, measure conversions, and enable cross-site retargeting

Device Fingerprinting

Collection of device and browser characteristics (OS, browser type, user agent, screen resolution, fonts, plugins) to create unique device identifiers for tracking across sessions and devices without explicit consent

Login-based Tracking

When users log into OpenAI accounts, their activity can be linked across devices and sessions, enabling cross-device profile building

Web Analytics

Analytics scripts embedded in OpenAI properties to collect usage data, session information, and user interaction patterns

Terms of Service Updates

Expanded restrictions on AI Output for consequential decisions

The new version explicitly prohibits using Output relating to persons for decisions with legal or material impact, including credit, educational, employment, housing, insurance, legal, and medical decisions. The previous version's language was cut off and incomplete, but the new version provides comprehensive coverage of high-stakes decision categories.

Effective: January 1, 2026

Added disclaimer about third-party endorsements and offensive content

New language clarifies that Output may be incomplete, incorrect, or offensive and does not represent OpenAI's views. Additionally, third-party product or service references in Output do not imply endorsement or affiliation. This is a clarification rather than a substantive rights change.

Effective: January 1, 2026

Updated URL format in links throughout document

All internal links were converted from relative paths (e.g., /policies/service-terms/) to absolute URLs (e.g., https://openai.com/policies/service-terms/). This is a technical/formatting change with no impact on user rights or obligations.

Effective: January 1, 2026

AI & Data Training

openai uses your content to train AI

Below is what's used and how (if at all) you can object.

What Content Is Used

  • Chat prompts and messages

    User-written prompts and conversation text used to train and improve AI models

  • Uploaded files

    Documents, code, and other files uploaded to OpenAI services used for model training

  • Images

    Images uploaded to or generated through OpenAI services used for model training

  • Audio recordings

    Audio files and voice input used for model training and improvement

  • Video content

    Video files uploaded to OpenAI services used for model training

How to Object

You can submit an objection using the steps below.

  1. 1

    Log into your OpenAI account, navigate to Settings > Privacy controls, locate 'Improve model training' or 'Use conversations to improve our models', and toggle OFF. This applies to new conversations after the change is made.

Why We Analyzed openai

TrueTerms automatically audits privacy policies and data practices using advanced machine learning to keep you informed and protected. This scorecard is based on the latest available public terms of service and privacy policies as of 2026.