Last checked:

Scorecard › Is ChatGPT safe for business?

Is ChatGPT safe for business?

Best-graded plan: ChatGPT Business (A). ChatGPT Business, ChatGPT Enterprise do not train on your data by default. ChatGPT Free, ChatGPT Plus can use your data for training unless you opt out.

PlanGradeTrains by defaultHuman reviewRetentionDPASOC 2
FreeCyes (opt-out available)unknownTemporary Chats up to 30 days; regular chats until you delete them, then removed within 30 days (unless already de-identified)not listed (OpenAI's DPA covers Business, Enterprise, API)not listed (OpenAI's SOC 2 Type 2 covers the API, ChatGPT Business/Team, Enterprise and Edu)
PlusCyes (opt-out available)unknownTemporary Chats up to 30 days; regular chats until you delete them, then removed within 30 days (unless already de-identified)not listed (OpenAI's DPA covers Business, Enterprise, API)not listed (OpenAI's SOC 2 Type 2 covers the API, ChatGPT Business/Team, Enterprise and Edu)
BusinessAnoLimited: authorized staff and contractors for support/abuseAdmin-controlled; deleted chats removed within 30 daysyesyes (Type 2)
EnterpriseAnoOnly for incidents, recovery with permission, or legal requirementAdmin-controlled; deleted chats removed within 30 daysyesyes (Type 2)

Sources: Free · Plus · Business · Enterprise

C ChatGPT Free

Trains on your chats unless you opt out. Thumbs-up/down feedback can still send the whole chat for training even after opting out. No business contract.

"When you use our services for individuals (Free, Plus, Pro), we may use your content to train our models."

C ChatGPT Plus

Same as Free: trains unless you opt out, with the feedback exception. No business contract.

"When you use our services for individuals (Free, Plus, Pro), we may use your content to train our models."

A ChatGPT Business

No training by default (feedback doesn't change that), admin-controlled retention, DPA, SOC 2 Type 2.

"We do not train our models on your data by default"

A ChatGPT Enterprise

No training by default, admin retention controls, SSO, DPA, SOC 2 Type 2.

"We do not train our models on your data by default"

Grades describe published policies, not security testing. How we grade.

More on ChatGPT

Read past issues of The Safe-to-Use Brief