Back to AI Privacy Guide

Midjourney: Privacy, Training & Output Ownership

Tier-by-tier analysis of Midjourney's data handling, training policies, and commercial output rights. Updated 2026-01-18.

Quick Answer

Midjourney is unsuitable for professional use involving sensitive or privileged information due to its public-by-default nature and lack of 'vault' protections. While it is a premier creative tool, it operates on a model that prioritizes community sharing and algorithmic training over individual data privacy.

Midjourney is a powerful creative tool but should never be used for confidential or proprietary work; its data practices are incompatible with legal or medical privacy requirements.

Privacy & Data Analysis

Sensitive Data

No

Used for Training

Yes

Output Ownership

Conditional

Sensitive Data

Midjourney is a public-by-default platform where prompts and images are visible to others and lack professional confidentiality protections.

Training

The service explicitly utilizes user prompts and uploaded images to train and refine its generative AI models.

Midjourney's Privacy Policy and Terms of Service (as of Jan 2026) confirm that data collected through the service is used for training its machine learning algorithms. This includes the text prompts provided by the user and any sample images or content uploaded to the service. There is no standard 'opt-out' for training on consumer-grade accounts, as the platform's development relies heavily on user-generated data to improve output quality.

Output Ownership

Users own the 'Assets' created on paid plans, but must grant Midjourney a perpetual, irrevocable license to use and reproduce them.

While Midjourney's terms state that paid subscribers own the 'Assets' they generate, this ownership is legally fragile. Users are required to grant Midjourney a 'perpetual, worldwide, non-exclusive, sublicensable, royalty-free, irrevocable' license to use, display, and distribute any generated content. Additionally, legal precedents in early 2026 continue to struggle with recognizing copyright for purely AI-generated works, meaning users may not have full intellectual property protection against third parties.

Data Retention

Data, including prompts and generated assets, is stored indefinitely to support the public gallery and training datasets. While users can delete images from the public view or use 'Stealth Mode' on high-tier plans to hide creations from the real-time feed, the underlying data often remains within Midjourney's internal systems for training and operational purposes.

Security Measures

Security is geared toward standard web platform protections rather than enterprise or regulated industry standards. The service lacks HIPAA compliance, SOC2 Type II certifications, and specific legal 'vault' doctrines. Its integration with Discord introduces additional third-party security risks, as data flows through a community chat platform before reaching Midjourney's servers.

Your Rights & Control

Users have the right to delete their accounts and request data deletion, but these rights are limited regarding data already incorporated into model training. Subscribers on Pro or Mega plans have access to 'Stealth Mode,' which prevents their images from appearing on the public Midjourney website, though the data is still accessible to Midjourney employees and systems.

Special Considerations

For attorneys and medical professionals, the use of Midjourney to visualize case details or patient scenarios is highly risky. Since the platform is considered a 'third party' without a strict confidentiality agreement (unlike a protected cloud storage provider), entering confidential info likely constitutes a waiver of professional privilege under current 2026 legal standards.

FAQ: Midjourney

Does Midjourney train on my inputs?

Midjourney: The service explicitly utilizes user prompts and uploaded images to train and refine its generative AI models. Midjourney's Privacy Policy and Terms of Service (as of Jan 2026) confirm that data collected through the service is used for training its machine learning algorithms. This includes the text prompts provided by the user and any sample images or content uploaded to the service. There is no standard 'opt-out' for training on consumer-grade accounts, as the platform's development relies heavily on user-generated data to improve output quality.

Can I use Midjourney with confidential or client data?

Midjourney: Midjourney is a public-by-default platform where prompts and images are visible to others and lack professional confidentiality protections.

Who owns the output I generate with Midjourney?

Midjourney: Users own the 'Assets' created on paid plans, but must grant Midjourney a perpetual, irrevocable license to use and reproduce them. While Midjourney's terms state that paid subscribers own the 'Assets' they generate, this ownership is legally fragile. Users are required to grant Midjourney a 'perpetual, worldwide, non-exclusive, sublicensable, royalty-free, irrevocable' license to use, display, and distribute any generated content. Additionally, legal precedents in early 2026 continue to struggle with recognizing copyright for purely AI-generated works, meaning users may not have full intellectual property protection against third parties.

What is Midjourney's data retention policy?

Midjourney: Data, including prompts and generated assets, is stored indefinitely to support the public gallery and training datasets. While users can delete images from the public view or use 'Stealth Mode' on high-tier plans to hide creations from the real-time feed, the underlying data often remains within Midjourney's internal systems for training and operational purposes.

Does Midjourney meet ABA Model Rule 1.6 confidentiality for lawyers handling client data?

No. Midjourney is not appropriate for client-confidential data under Model Rule 1.6. Use an enterprise tier of a different service or an explicit no-training tier of Midjourney if available. See the AI Privacy Guide at https://hoaglaw.ai/resources/ai-privacy-guide for the full comparison.

Need an AI-aware contract review or governance policy?

Hoag Law.ai builds AI-aware MSAs, DPAs, and internal governance frameworks for startups, flat-rate from $2,500/month. If you're evaluating Midjourney for your team, let's talk.

Book a free call

This site uses cookies for live chat (Crisp). No analytics cookies. Privacy details