AI-Specific Policiesยท4 min readยทEffective 1 April 2026

AI Content Restrictions

What AskBiz AI will and will not generate. Hard limits, content the AI will decline, and how to report an AI response that violated these restrictions.

Last updated: 1 April 2026 ยท Questions? legal@askbiz.co

Hard Limits โ€” What the AI Will Never Generate

Regardless of how a request is framed, AskBiz AI will never generate:

  • Content that facilitates, assists with, or encourages illegal activity
  • False or fabricated financial data intended to deceive any third party
  • Content that assists with fraud, money laundering, or tax evasion
  • Advice on circumventing legal or regulatory requirements
  • Content that violates sanctions or export control law
  • Defamatory, harassing, or discriminatory content about individuals or groups
  • Content that exploits, harms, or endangers minors in any way
  • Detailed instructions for harmful, dangerous, or illegal activities

These limits are implemented at both the system level (via Anthropic's Claude model safeguards) and at the AskBiz application level.

Business-Scope Restrictions

AskBiz AI is scoped for business intelligence. It will decline requests that fall outside this purpose:

  • Personal advice unrelated to business (medical, legal, relationship advice)
  • Content creation unrelated to business analysis (creative writing, general-purpose text generation)
  • Political commentary or advocacy
  • Requests to roleplay as a different AI system or remove its guidelines
  • Requests to generate content in bulk for SEO manipulation or content farming

If you have a legitimate business need that the AI is declining, contact support@askbiz.co. We review refusal patterns and adjust where appropriate.

How Restrictions Are Enforced

Content restrictions are enforced at two levels:

Model level: Anthropic's Claude model has built-in safety guidelines that cannot be overridden by AskBiz or by users. These represent Anthropic's responsible AI commitments.

Application level: AskBiz applies additional restrictions appropriate to a business analytics context. These are implemented via system prompts and output filtering.

Neither level is infallible. If you encounter an output that appears to violate these restrictions, use the thumbs-down feedback mechanism immediately and email security@askbiz.co with the conversation context.

What to Do When the AI Declines a Legitimate Request

If the AI declines a request you believe is legitimate:

1. Rephrase the question โ€” sometimes the framing triggers a refusal. Try asking more specifically or providing more context.

2. Check the scope โ€” AskBiz AI is focused on business analytics. If your request is outside this scope, it may be declined by design.

3. Contact support โ€” email support@askbiz.co with the context. We review refusal logs and will help where the refusal was incorrect.

We log all refusals (not the content of your request, just the fact of refusal and general category) to improve the system over time.