Midjourney runs every prompt through an automated moderator before it reaches the model. A block is not a bug.
Why prompts get flagged
- Banned words, including medical and anatomical terms used innocently.
- Named public figures in certain contexts.
- Gore, explicit content, or violence descriptors.
- Attempts to circumvent the filter with spacing, symbols or misspellings — these escalate faster than the original term would.
Rewriting a flagged prompt
- Replace the flagged term with a visual description of the same idea. "Blood" becomes "deep crimson liquid".
- Remove intensifiers such as "brutal" or "graphic" — they trip the classifier independently.
- Describe the composition and lighting instead of the subject's state.
- Never obfuscate with
b l o o dorbl00d. The filter catches it and it counts as an evasion attempt.
Warnings and bans
Repeated flags produce warnings, then a temporary suspension, then a permanent ban. If you believe a block was a false positive, appeal through your Midjourney account page rather than repeating the prompt.
Silent failures
Sometimes a flagged job simply never returns instead of showing an error. If jobs vanish with no message, see Midjourney not loading? Fix stuck jobs fast.