AvocadoAI Content Policy
This content policy explains how creators can safely use AvocadoAI’s image, video, music, voice, and editing tools.
AvocadoAI Content Policy
Last updated: November 2025
Purpose
AvocadoAI lets creators and brands generate professional images, videos, voice, and music using world-class AI models.
To keep the platform safe, legal, and fair, every generation must follow this policy.
This page explains what may trigger content filters and how to avoid violations when using models like Veo 3.1, Sora 2, Imagen 4, SeedDream v4, Nano Banana, and others.
Overview
All models automatically scan your prompts and outputs for restricted content.
If a generation is blocked, you may see messages such as:
- “The content could not be processed because it contained flagged material.”
- “Content policy violation: rejected by content filters.”
- “Unexpected value; permitted: ‘9:16’, ‘16:9’"
These messages mean your prompt or reference content breached a safety rule, not that the system failed.
Global Restrictions (Apply to All Models)
You may not use AvocadoAI to generate or edit content that:
Sexual or Adult Content
- Pornography, sexual acts, explicit nudity, or fetish imagery
- Minors or individuals appearing under 18 in any sexual or suggestive context
- Non-consensual or exploitative sexual situations
Violence and Gore
- Realistic depictions of injury, blood, death, or torture
- Content intended to shock, disgust, or glorify violence
Hate or Extremism
- Promotion of hate groups or extremist ideology
- Symbols, gestures, or language targeting protected classes
Misinformation and Deepfakes
- False or misleading representations of real events or people
- Political or election manipulation videos or images
- Deepfakes of real individuals without consent
Illegal or Unsafe Activity
- Drugs, weapons, or criminal activity
- Self-harm, suicide, or eating-disorder promotion
Copyright and Trademark
- Recognizable characters, logos, or brands (Disney, Marvel, Nike, etc.)
- Music or voices mimicking known artists or copyrighted works
Privacy Violations
- Real people’s faces, voices, or names without permission
- Personal data, identification cards, or private information
Medical or Financial Misuse
- False health claims, medical diagnostics, or financial advice imagery
Model-Specific Rules
Image Generation (Imagen 4, SeedDream v4, Flux, Nano Banana)
Avoid:
- Realistic sexual scenes or partial nudity
- Graphic injuries, warfare, or weapons
- Real people’s faces or names
- Logos, movie characters, or branded objects
- Political propaganda posters
Recommended:
- Specify style or medium: “digital art,” “studio photo,” “oil painting,” “3D render”
- Describe fiction or abstraction: “in a fantasy city,” “concept artwork,” “training poster style”
- Use roles instead of names: “a journalist,” “a superhero,” “a woman in business attire”
Image Editing (Nano Banana Edit, Flux Kontext)
When editing or inpainting:
- Do not upload or reference real photos of identifiable individuals unless you have consent.
- Do not attempt “nude reconstruction,” “clothing removal,” or “real-person swaps.”
- Keep edits artistic or compositional: lighting, color, background, pose, or style changes.
- Avoid inserting weapons, hate symbols, or explicit material.
Tip: describe intent clearly, e.g. “Replace background with tropical beach, same lighting” or “Adjust colors for warmer tone.”
Music Generation (Minimax Music v2, ElevenLabs Music)
Prohibited:
- Lyrics or titles containing hate speech, slurs, or sexual content
- Music imitating a known artist’s voice, melody, or copyrighted composition
- Political jingles or propaganda pieces
Guidelines: - Describe mood, tempo, and instrumentation: “lo-fi hip-hop with soft piano,” “energetic EDM intro”
- Keep lyrics neutral or original.
- Do not upload copyrighted lyrics for continuation or remix tasks.
Voice Generation (Avocado Voice, ElevenLabs Voice)
Strictly forbidden:
- Cloning or mimicking real people without consent
- Fake news recordings, scam calls, or misleading narration
- Sexual, violent, or hateful voice content
Safe usage: - Use fictional personas or roles: “friendly narrator,” “animated robot guide,” “female teacher”
- Mention tone and emotion: “warm and calm,” “excited,” “formal”
- If referencing a celebrity, rephrase to style only: “in a confident movie-trailer style,” not “in Morgan Freeman’s voice.”
Video Generation (Veo 3.1, Sora 2)
Video models are the most regulated due to realism and potential misuse.
Forbidden in Video Prompts:
- Realistic depictions of real people (celebrities, politicians, influencers)
- Any sexual, violent, or exploitative acts
- Scenes of war, disaster, or crime presented as factual
- Political content, campaign messages, or election scenarios
- Deepfake use (recreating existing footage or merging real people)
- Brand/IP scenes using known characters, logos, or product packaging
- Footage resembling real news (anchors, press conferences, reports)
Recommended Practices:
- Always clarify fiction: “In a fictional city,” “short animated film,” “stylised cinematic trailer.”
- Describe setting and tone: “Cinematic shot, soft light, slow motion, peaceful mood.”
- Use generic roles: “a scientist,” “a hero,” “a child reading.”
- Define style: “3D animation,” “digital art,” “cartoon,” “surreal cinematic.”
- Limit realism when uncertain: Prefer “stylised” or “animated” over “photo-realistic.”
- Keep duration short (≤ 8 seconds) and ratio valid (9:16 or 16:9).
Veo 3.1 Notes
- Audio and video generations are heavily filtered.
- Realistic dialogue or lip-sync with named persons will be blocked.
- Avoid prompts implying documentary or real recording.
Sora 2 Notes
- Strict on realism, likeness, and event references.
- Avoid prompts referencing “real footage,” “breaking news,” “leaked video,” or any sexual or violent act.
- Use creative framing: “dream sequence,” “abstract short,” “cinematic test scene.”
Safe Prompt Examples
Goal | Unsafe Prompt | Safe Prompt
Realism | “Video of Elon Musk giving a speech at a rally” | “Animated tech leader presenting at a futuristic conference, fictional city”
Action | “Explosion in downtown LA caught on camera” | “Stylised sci-fi explosion on alien planet, cinematic lighting”
Portrait | “Photo of Taylor Swift at party” | “Studio portrait of a female musician in neon lighting, no specific identity”
Music | “Song in style of Drake’s God’s Plan” | “Modern hip-hop track with deep bass and confident flow”
Voice | “Make voice identical to Barack Obama” | “Warm authoritative male narrator tone”
Edit | “Remove clothes from photo” | “Adjust lighting and background for neutral portrait”
Filtering and Enforcement
How Filters Work
- Prompts are scanned for restricted keywords or descriptions.
- Outputs are analyzed for nudity, violence, or realistic likeness.
- Flagged generations are automatically blocked.
User Response
When blocked:
Review your prompt and remove restricted elements.
Use fictional, stylised, or generic wording.
Regenerate with safer phrasing.
Repeated violations may result in:
- Temporary suspension of generation access
- Permanent account ban for repeated or malicious use
Reporting and Appeals
If you believe your prompt was incorrectly flagged:
Save your prompt and timestamp.
Contact hello@avocadoai.co with subject line “Content Policy Appeal.”
Include a short explanation of your creative intent.
The team reviews cases manually to refine filter accuracy.
Transparency and User Responsibility
All creators are responsible for the content they produce and share.
Generated media must not be presented as real footage or factual recordings.
When sharing public output, include a disclosure such as “Created with AvocadoAI.”
Updates
This policy may change as new models and regulations emerge.
Significant updates will appear on the AvocadoAI Knowledge Base and in user notifications.
Summary Checklist
Before generating, ask yourself:
- Does my prompt mention real people, brands, or events?
- Could this content be misinterpreted as real footage?
- Is there any sexual, violent, or hateful context?
- Did I specify style and fiction clearly?
If any answer is “yes,” adjust the prompt before generating.
Contact
For questions or clarification:
Email: hello@avocadoai.co
Website: https://avocadoai.co