Content Risk Review Rule Types

IntelligenceBank's Content Risk Review is a feature that scans and flags brand, legal, and/or compliance-related risks identified within content uploaded to the IntelligenceBank platform. These risks are identified based on underlying rules that are configured according to your organisation's regulatory requirements, with the help of the IntelligenceBank rules team.

There are several different types of content risk review rules that your organization can use to help detect risks within your content. 

IntelligenceBank offers many AI services and products that align with your organization's individual requirements and adhere to your internal AI governance and usage policies. For content risk reviews, we offer the following options. We can also tailor subsets of rules to specific methodologies, e.g. if you have a preference for disclaimer rules to only be created with deterministic AI and a preference for LLM-based rules to identify brand guidelines risk:

Traditional Machine Learning (ML)-based rules (Deterministic) 

• Includes rule-based AI
• Rules are developed using only traditional machine learning techniques; No generative AI.
• These rules are generally best used when checking for specific, known risks, such as clear yes/no decisions, non-negotiable requirements, or fact-based checks.
  
  

Hybrid ML-LLM-based rules

• Source of Truth rules – These are used for long-form documents that contain many requirements. The AI reviews the source content(s) and ensures that the uploaded material is aligned with those requirements.

Large Language Model (LLM)-based rules (Non-Deterministic)

• Advanced AI rules – Our specialised rule consultancy team designs and tailors a solution that reflects a requirement or a set of requirements, and the AI applies it to the uploaded content by tapping an LLM model through our API.
• These rules are generally used when there's a "gray area" or judgment involved in determining whether there's a risk, catching nuances that deterministic rules may miss, evolving or ambiguous scenarios, or brand tone, sentiment, or messaging risks.
  
  
• Agentic rules – These are out-of-the-box (OOTB) rules designed by our specialised rules consultancy team based on specific regulatory or other best practices, such as:
• Over promissory language
• Greenwashing language
• Appeal to Minors language
• General Advisory language
• Spelling & Grammar