Witivio Documentation

# GPT Pro Content Filtering

GPT Pro includes built-in content filtering so conversations stay safe and appropriate for the workplace. Filtering works in two directions: it declines requests on sensitive topics, and it cleans every reply before it reaches your Microsoft Teams chat. This protection is always on and cannot be turned off.

# Sensitive topics GPT Pro declines

To keep the assistant suitable for professional use, GPT Pro will not engage with restricted subjects. When a request involves one of these topics in a non-educational way, GPT Pro replies with a short, polite refusal rather than an answer.

Restricted topics include:

  • War and weapons
  • Drugs
  • Medical advice
  • Sexual advice
  • Politics
  • Religion
If GPT Pro cannot help, it responds with a message such as I am sorry, but I cannot answer this question. Rephrase your request or focus on a work-related task to continue.

# Safe, sanitized responses

Every GPT Pro reply is generated as clean, sanitized HTML before it appears in Microsoft Teams. This keeps responses readable and blocks anything that could be unsafe.

  • Only safe formatting tags are used, such as paragraphs, lists, bold, and italics.
  • Scripts, styles, inline frames, and event attributes are never included.
  • Links using unsafe protocols are blocked.
  • If you ask GPT Pro to embed raw code or scripts, it politely declines for security reasons.
Because replies are always sanitized, you can safely copy text from GPT Pro into your messages, emails, or documents without carrying over hidden or unsafe content.

# Privacy and data handling

Content filtering works alongside strong data protection. GPT Pro follows enterprise standards for how your information is handled.

  • Hosted in Microsoft Azure in Europe.
  • Fully GDPR compliant.
  • Your data is never used to train external models.
  • No access to your Microsoft Teams messages, files, or corporate systems.
  • Only the text you type into the chat is processed.

Retention and governance follow your organization's Microsoft 365 policies.

← What's new →