Introducation:
Banned words Midjourney have caused Midjourney’s AI to reject certain image generation requests. Midjourney is an AI art generator created by Anthropic. It has gained great popularity since its launch in 2022, and with the increased use of its powerful image-generation capabilities, there have also been several misuses. The team behind the powerful AI image generator has had to make several changes to prevent users from generating images that could potentially be inappropriate, offensive, or even legally questionable. Midjourney aims to maintain a PG-13 rating or lower.
Some general topics that are restricted include graphic depictions of violence, nudity, criminal plans or acts, and hate speech targeting characteristics like race, religion, gender, or sexual orientation. Understandably, Midjourney also bans direct references to copyrighted characters or logos without permission.
Midjourney has policies in place around certain types of prompts and generated images to prevent harmful, dangerous, or illegal content. Here are some key things to know about Midjourney banned prompts:
- Crude Phrases:
Midjourney will immediately detect banned words within any prompt containing swears, slurs, or derogatory terms and Midjourney will not generate any image.
- Explicit Content:
Midjourney is strictly against any NSFW content so words relating to nsfw themes, genitalia, or pornography will also be rejected immediately. So, trying to create Midjourney porn would be impossible.
- Lack of Attire:
Any prompts that could generate an image of a nude person/character are also strictly forbidden. Prompts with implied nudity are also among the banned words in Midjourney AI. Implied nudity in Midjourney refers to prompts that suggest or indirectly involve nude portrayals without explicitly describing banned sexual content.
Some examples that may cross the line into implied Midjourney nudity:
- A nude figure with strategic positioning/posing to conceal explicit details.
- Requests for artwork showing intimate areas that are partially obscured.
- Omission of certain limbs/anatomy while still making the nudity context obvious.
- Descriptions of scenarios that imply participants are nude (e.g. in a shower) without saying so directly.
- Focusing prompts on depicting bare skin or specific areas of the body often associated with sex appeal.
- Overly provocative poses that sexualize the subject despite technical coverage.
- Violence/Gore:
Prompts suggesting graphic depictions such as harm, injury, gore, abuse, etc. Although in some cases one can argue that this kind of imagery can be used for educational purposes, it is still strictly forbidden. Mild fantasy violence in realistic art styles is typically okay, but gratuitous or disturbing violence crosses the line.
- Illegal Acts:
Midjourney will not generate any images portraying illegal acts like vandalism, assault, drugs, etc.
- Hate Speech:
Racism, discrimination, and harmful stereotypes are also among the banned words in Midjourney.
- Copyrighted Works:
Copyrighted characters, artworks, or brands typically cannot be requested without permission. According to Midjourney Creative Commons licensing Midjourney retains full copyright over all images it produces. Directly referencing copyrighted works without permission goes against their policy as it could unintentionally cause a copyright infringement.
- Politically Sensitive Topics:
Explicitly using Midjourney’s capabilities to spread hate, intolerance, or promote discrimination against others would not be allowed. This includes:
- Propaganda against protected attributes like race, gender, etc.
- Extremist or terroristic imagery/messaging
- Symbols or phrases intended to provoke or incite violence
Political and religious art may be generated.
How to Bypass the Midjourney NSFW filter?
Directly bypassing Midjourney’s NSFW filter is not possible as all the generated images are scanned by the AI before being presented. So even if you somehow bypass prompt restrictions, actually getting the image is another story. You can, however, create a Midjourney NSFW workaround. Beyond specific banned terms, context is key. A word may be permitted if used metaphorically versus a literal description. For example, mentioning a historical war in a historical context presents less risk than glorifying violence. Similarly, artistic nudity done respectfully differs from sexually-charged content. Modifiers like “concept art”, “fictional”, or descriptions instead of direct terms may also make problematic prompts acceptable but there are no guarantees.
There have been reports of prompts slipping through initially but then getting banned retroactively after review. Users can have their accounts suspended or banned if found to be consistently requesting prohibited prompts or sharing banned images. The AI itself can’t understand context like humans, so it plays it safe. Customer service can sometimes overrule automatic bans, so keep that in mind if you ever get banned.
To avoid having your prompt rejected or unintentionally violating the community guidelines you can use the Midjourney negative prompts section to blacklist these prompts so that Midjourney will explicitly avoid using them. While certain words understandably remain private, the goal of the filters is to sidestep unintended offensive, unethical, or illegal outputs from an AI without boundaries. With mindful wording that considers broader impacts, creative users can still explore many conceptual themes.
It’s also worth noting that although Midjourney strictly forbids certain prompts and images by filtering them, it may not always be able to filter everything and might generate images that go against its own guidelines.
