Are your Midjourney prompts and images private? On the lower tiers they are published to other users by default — including anything you typed into them.
Plain-English risk rating: 4 of 5 (Basic/Standard) / 3 of 5 (Pro/Mega with Stealth Mode)
Midjourney sits high on the consumer-AI risk axis for a reason most users do not understand: everything you create on the lower tiers is public by default. Your prompts and your generated images are visible to other Midjourney users on the web gallery and across Discord. This is structurally different from every other vendor in this database — ChatGPT, Claude, Gemini, and so on retain your conversations on their servers but do not expose them to other users. Midjourney's default is exposure.
The second material risk: Midjourney is the defendant in an ongoing Disney/Universal copyright lawsuit (Disney Enterprises Inc. et al v. Midjourney Inc., C.D. Cal. No. 2:25-cv-05275, filed 11 June 2025; no trial date has been set) alleging training on copyrighted character images. The outcome could fundamentally alter commercial use policies. Combined with the February 2026 Supreme Court Thaler decision confirming AI-generated works cannot be copyrighted, the IP picture for businesses using Midjourney commercially is unsettled.
The Pro and Mega tiers with Stealth Mode enabled bring the risk down to 3 of 5 by making your prompts and images private, but Stealth Mode does not delete already-generated public content and does not change Discord's separate data retention.
Recommended for
- Sole proprietor designer: Standard Plan at $30/month is the minimum we would suggest for commercial work, but it still exposes your prompts publicly, with no option of Stealth Mode. For sensitive client work, upgrade to Pro at $60/month for Stealth Mode.
- Small team (2-10 people): Pro or Mega with Stealth Mode enabled. Document your commercial usage rights internally and audit which images on your team's gallery are still publicly visible from earlier generations.
- Regulated industry: No general guidance — Midjourney is not architected for regulated-content workflows. Consider Adobe Firefly Enterprise instead (separate profile in this database).
- The honest answer for most 1-10 employee businesses using AI image generation commercially: Midjourney produces higher-aesthetic-quality output than most competitors for most prompts as of mid-2026, which is why it remains the dominant choice despite the privacy issues. The right configuration for commercial use is Pro or higher with Stealth Mode permanently enabled. Treat any prompt you write on a lower tier as published material.
Critical pre-deployment warning (revenue threshold)
Midjourney's Terms of Service contain a revenue-trigger clause: if you (or your company) generate more than $1,000,000 USD per year in revenue, you MUST be subscribed to Pro or Mega to own your generated Assets. Smaller businesses on lower-tier plans retain ownership; larger businesses on lower-tier plans do not. This is a structural compliance issue for any SMB on a growth trajectory — your subscription tier needs to match your revenue tier or you risk losing ownership of work you have built into your products.
Data retention default
- Basic, Standard: All prompts and generated images public by default on the Midjourney web gallery and Discord servers. Visible to millions of other users. Midjourney publishes no retention period for content you do not delete, so it stays in the public archive until you delete it. Its Data Deletion and Privacy FAQ (updated 3 September 2026) says deletion completes within 30 days, with a 7-day window to reverse it. (Corrected 2 October 2026: this previously said images are retained indefinitely.)
- Pro, Mega with Stealth Mode ON: New generations are private from this point forward. Existing public generations from earlier sessions are NOT retroactively privatised — they remain visible. Some prompt-level metadata may still be retained on Midjourney's systems.
- Discord retention: Whether you generate via Midjourney's web app or via Discord, Discord's own data retention applies in parallel. Discord retains messages, server activity, and metadata under its own privacy policy, separate from Midjourney's.
Training opt-out
TRAINING IS ON BY DEFAULT and there is no per-user opt-out for the consumer tiers. Midjourney trains on user-generated images and prompts as part of its model improvement loop.
Midjourney's own AB2013 documentation discloses that the training dataset comprises:
- Publicly available data crawled from the public web and public repositories
- Internally generated data
- User interactions and inputs to the platform
- Licensed data and data that may be protected by copyright (the company asserts fair use)
For users in California, the AB2013 disclosure documentation provides additional transparency but does not provide an opt-out mechanism. Enterprise customers can negotiate custom terms.
Zero Data Retention availability
- Not offered
- Stealth Mode is the closest equivalent for prompt/image privacy, but it is a visibility control rather than a retention control — Midjourney still retains the data on its servers
Plan tiers and pricing (as of early 2026)
| Tier | Price (USD) | Default visibility | Commercial rights | Suitable for |
|---|---|---|---|---|
| Basic | $10/month | Public | Yes (under revenue threshold) | Hobbyists; not appropriate for client work |
| Standard | $30/month | Public | Yes (under revenue threshold) | Minimum acceptable for commercial work; still publicly visible |
| Pro | $60/month | Stealth Mode available | Yes; required for >$1M revenue companies | Commercial work requiring privacy |
| Mega | $120/month | Stealth Mode available | Yes; required for >$1M revenue companies | Heavy commercial use |
Plan detail re-checked against Midjourney own documentation on 3 September 2026, and two things needed correcting. Midjourney no longer lists a free tier or trial. Its current plan documentation shows four paid plans only: Basic $10, Standard $30, Pro $60 and Mega $120 a month, with a 20 percent discount on annual billing. The prices above are correct; the Free / Trial row was removed on 2 October 2026 because it is not offered and the rating label has been changed from Free/Basic/Standard to Basic/Standard. And a limitation on Stealth Mode that this profile did not carry at all: Midjourney states that even with Stealth Mode on, "any creations made in public channels on Discord are still visible to others". Paying for Pro or Mega does not make you private if you generate in a public Discord channel. To actually be private you must generate on the web Create page, in a Discord direct message, or in a private Discord server, with Stealth Mode on.
Jurisdiction
- Primary processor: Midjourney, Inc., San Francisco, California, USA
- Cloud infrastructure: Not publicly disclosed in primary documentation
- Data processed in the United States by default
- Subject to California privacy law (CCPA/CPRA) and California AB 2013 generative AI training data disclosure requirements
- Discord-based interactions subject to Discord's separate privacy policy and jurisdiction
Breach history (public incidents and ongoing legal proceedings)
June 2025 — Disney and Universal v. Midjourney (UNRESOLVED)
The Walt Disney Company and Universal Studios filed a federal copyright infringement lawsuit against Midjourney alleging the AI image generator trained on copyrighted character images and produces outputs that reproduce protected characters (Mickey Mouse, Spider-Man, the Simpsons, and others) on request. Corrected 26 September 2026: this section previously said a trial date was set for late 2026. No trial date has been set. Read on the court docket record on 26 September 2026: the case is Disney Enterprises Inc. et al v. Midjourney Inc., United States District Court for the Central District of California, No. 2:25-cv-05275, before Judge John A. Kronstadt, filed 11 June 2025, pleaded as copyright infringement under 17 U.S.C. 501. No ruling on any motion to dismiss has issued, no merits ruling has issued, and the case is in discovery. The last day to file all motions is 23 November 2026 and the court has estimated a 14-day trial, so a trial realistically falls in 2027. The Warner Bros. action was filed separately as No. 2:25-cv-08376 and consolidated with this case on 4 November 2025, so Warner Bros. is now part of the same proceeding. The parties were ordered into private mediation, with a post-mediation status conference on 31 August 2026. What this means for our rating: our rubric moves a score by 3 points only once litigation has survived a motion to dismiss, and by 1 point where material litigation exists below that line. No such ruling exists here, so this is a 1-point matter, and we say so rather than implying a court has tested these claims. If the plaintiffs prevail, the result could fundamentally alter commercial use policies for Midjourney across all tiers.
Sources: the court docket record for C.D. Cal. No. 2:25-cv-05275, read 26 September 2026; Terms.Law analysis (January 2026, updated March 2026). Corrected 26 September 2026 — this line previously credited “Court filings”, which implied we had read the docket. We had not, and the trial-date error above is what that produced.
February 2026 — Supreme Court declines Thaler petition (precedent affecting all Midjourney commercial use)
The US Supreme Court declined to hear Thaler v. Perlmutter, leaving in place the lower-court ruling that pure AI-generated works cannot be copyrighted under US law because they lack human authorship. The practical implication for Midjourney users: while you can use Midjourney outputs commercially per Midjourney's license, you cannot stop others from copying those outputs unless you have added enough human creative input to meet the human-authorship threshold.
Source: US Supreme Court order list, February 2026
July 2026 — Bartz v. Anthropic $1.5B settlement approved (precedent affecting AI training data lawsuits)
While this settlement involved Anthropic and text training data rather than Midjourney and image training data, the $1.5 billion settlement (about $3,000 per work for 482,460 works, final court approval 20 July 2026) establishes a precedent that could apply to image training data cases including the Disney/Midjourney lawsuit.
Source: Pearl Cohen, "Federal court approves $1.5 billion Anthropic copyright settlement" (read 23 October 2026). Corrected 23 October 2026: this entry was dated March 2026 and gave the settlement as about $3,000 per work for 7 million books, which would be $21 billion; the trial line above said late 2026, which this profile's own 26 September correction had already withdrawn.
Ongoing — Default-public visibility as a structural data exposure
Not a discrete breach incident, but a persistent structural issue worth treating in this section: every prompt typed on a Basic or Standard tier account is visible to every other Midjourney user via the web gallery. This includes client briefings paraphrased into prompts, internal project descriptions, competitive analysis terms, and any other text that an SMB owner might type when generating commercial work. The Privacy Watchdog score for Midjourney is 38/100 (D+), the lowest among major AI services, primarily driven by this default-public posture.
Source: Privacy Watchdog analysis (January 2026)
What this means in plain English for SMB owners
Three honest takeaways:
- Every prompt you type on Standard or below is published to other Midjourney users. This is not a bug; it is the documented default. If you have ever typed a client name, a confidential project description, or a competitive observation into a Midjourney prompt on a non-Stealth-Mode tier, that text is in the public Midjourney gallery and indexed by Discord's own systems.
- The Disney/Universal lawsuit could materially change Midjourney's commercial terms. No trial date has been set. The outcome is genuinely uncertain. SMB owners building business processes around Midjourney should have a contingency plan if commercial use rights are restricted after judgment — Adobe Firefly is the most credible alternative for that scenario because Adobe trained Firefly only on licensed and public-domain content.
- If your company is approaching $1M annual revenue, your Midjourney subscription tier needs to track that. The Terms of Service revenue-trigger clause means companies above $1M USD revenue must be on Pro or Mega to own their generated Assets. This is a compliance issue most growing businesses are not tracking.
Sources
- Midjourney Terms of Service: https://docs.midjourney.com/hc/en-us/articles/32083055291277-Terms-of-Service (verified 2026-05-24)
- Midjourney Privacy Policy: https://docs.midjourney.com/hc/en-us/articles/32083472637453-Privacy-Policy (verified 2026-05-24)
- Midjourney AB2013 Documentation (California training data disclosure)
- Terms.Law: Midjourney Commercial Use Rights March 2026 guide
- Terms.Law Privacy Watchdog: Midjourney Privacy Review (Score 38/100)
- Cold Iron Labs: Understanding The Midjourney Privacy Policy (November 2025)
- US Supreme Court order list, February 2026 (Thaler v. Perlmutter cert denial)
- Disney/Universal v. Midjourney case filings (June 2025, court docket)
Related on AI Leakage
- Compare all 29 AI tools in the risk directory — see how Midjourney stacks up against the rest, tier by tier.
- Take the 5-minute “Am I Leaking?” check — a personalised view of your business’s AI exposure.
- Check a prompt before you paste it — our free Data-Safe Prompt Rewriter.
- Shadow AI vs AI leakage — why even approved tools like Midjourney can leak data.
- Get plain-English AI Leakage Alerts — we email you when an AI tool you use changes its data policy or has an incident.
- Get the free AI Acceptable Use Policy template — a plain-English policy with the tool-by-tool risk guide built in.
Compare similar tools: Midjourney is one of 6 tools in AI writing, design and voice tools compared, which sets out each one’s rating and what it does with your data, side by side.