AI Safety Standards

Every Acorevia AI AI model must clearly identify itself as an AI model trained on approved materials. It must not claim to be the real person, impersonate a human, or mislead users about its nature.

Acceptable disclosure: "I am the official AI model trained on approved content from [Creator Name]."

Not permitted: Any statement that implies the model is the actual person, a human, or a live representative.

AI models may only be trained on materials submitted by the client and confirmed as owned, licensed, or legally permitted for use.

Acorevia AI does not scrape the internet, use third-party content without permission, or incorporate materials beyond those explicitly approved by the client.

Clients define the scope of topics the AI model should address. Our team configures the model to refuse questions outside that scope rather than speculating or hallucinating answers.

Default restricted topics for all models include:

• Legal advice
• Medical advice
• Financial advice
• Political opinions
• Private or confidential personal matters
• Content that could cause harm to users

Additional restrictions are configured at client request during the build process.

Where an AI model operates in areas adjacent to professional advice (health, finance, law, therapy, spiritual counselling), appropriate disclaimers are configured and displayed.

These disclaimers direct users to seek qualified professional guidance where appropriate.

No Acorevia AI AI model is launched publicly without completing a private testing phase. During testing:

• The client receives access to test the AI model.
• Flagged responses are reviewed and corrected by our team.
• The client provides explicit written approval before launch.

Acorevia AI reserves the right to refuse launch of any AI model that does not meet safety standards, regardless of client pressure.

Public launch is locked until all of the following conditions are met:

• Private testing is complete.
• All flagged issues are resolved.
• The client has given explicit written approval.
• Final payment has been received.

Acorevia AI recommends ongoing monitoring of AI model responses after launch. Post-launch monitoring and maintenance packages are available for all tiers.

Clients are responsible for monitoring use of their AI model after delivery and reporting any safety concerns to Acorevia AI promptly.

Acorevia AI will not build AI models intended to:

• Deceive or manipulate users.
• Produce or distribute harmful, illegal, or unethical content.
• Impersonate real individuals in misleading ways.
• Operate without AI identity disclosure.
• Violate applicable laws, regulations, or platform policies.

Acorevia AI reserves the right to update these Safety Standards as AI technology and best practices evolve. Active clients will be notified of material changes. Existing AI models may need to be updated to remain compliant.