OpenAI's data handling commitments
OpenAI has made specific commitments about how writing style profile data is handled. These commitments are different from how standard ChatGPT conversations are handled.
| Commitment | Standard ChatGPT | Writing style feature |
|---|
| Used for model training? | Yes (unless opted out) | No |
| Stored on OpenAI servers? | Yes, for 30 days | Yes, for duration of feature activation |
| Reviewable by OpenAI staff? | Possible | No (encrypted, automated only) |
| Deletable by user? | Yes (per conversation) | Yes (entire profile) |
| Subject to legal process? | Yes | Yes (subject to legal process) |
Source: OpenAI privacy disclosure, September 2026; OpenAI Help Center data handling, September 2026.
Comparison to competitors
OpenAI's ChatGPT is the first to offer comprehensive personal context + writing style mimicry in a single product. Competitors offer pieces but not the full combination.
| Capability | ChatGPT (OpenAI) | Claude (Anthropic) | Gemini (Google) | Copilot (Microsoft) |
|---|
| Personal app integration | Gmail, Calendar, Notion, Drive | Limited (Project, Artifacts) | Workspace integration | Microsoft 365 integration |
| Writing style mimicry | Full profile extraction | Limited (Projects) | Limited (Workspace context) | None |
| Cross-app context | Yes | Partial | Yes (Workspace only) | Yes (Microsoft 365) |
| Privacy controls | Disable per integration | Project-based | Workspace admin | Tenant admin |
Source: OpenAI vs competitors feature comparison, September 2026; ChatGPT vs Claude vs Gemini 2026; Copilot feature documentation, September 2026.
Enterprise and organizational considerations
Organizations should evaluate whether employee use of this feature violates data handling policies, particularly for industries with strict data residency requirements. The feature reads email content that often contains sensitive business information.
For organizations in regulated industries (healthcare, finance, government contracting), employee use of the ChatGPT writing style feature may violate data handling policies if it transmits PHI, PII, or other regulated data to OpenAI servers. Enterprise administrators should: (1) review current data handling policies for AI tool usage, (2) determine whether to enable or disable the feature at the organization level, (3) train employees on appropriate use, (4) monitor for policy violations through DLP (Data Loss Prevention) tools. The ChatGPT Enterprise plan offers organization-level controls that can disable the feature entirely or limit which integrations are allowed (OpenAI Enterprise documentation, September 2026; Gartner AI governance guidance, September 2026).
How to enable and disable the feature
Users can enable or disable the personal apps feature at any time through ChatGPT settings. Disabling removes the writing style profile and stops future reads.
| Step | Action | Notes |
|---|
| 1. Open ChatGPT settings | Click profile icon > Settings | Web or desktop app |
| 2. Navigate to Personal apps | Settings > Personalization > Personal apps | List of connected apps |
| 3. Connect a new app | Click Connect, follow OAuth flow | Permissions explained |
| 4. Review writing style profile | Settings > Personalization > Writing style | See extracted patterns |
| 5. Delete profile | Click Delete writing style profile | Removes all extracted data |
| 6. Disable feature | Toggle off Personal apps | Disconnects all apps, deletes profile |
Source: OpenAI Help Center writing style guide, September 2026; OpenAI support documentation, September 2026.
FAQs
The questions above cover what the personal apps feature is, how the writing style mimicry works, which apps integrate, privacy implications, which plans include it, and how it compares to competitors.