Content Risk Review Rule Types

At IntelligenceBank, we are proud to offer a multitude of AI services and products that align with individual customer requirements and adhere to their internal AI governance and usage policies. For content risk reviews, e.g. brand, legal and compliance rule risk detection, we offer the following options for clients. We can tailor subsets of rules to specific methodologies, e.g. preference for disclaimer rules to only be created with deterministic AI and preference for LLM-based rules to identify brand guidelines risk:

Traditional ML-based rules (Deterministic) 

• Includes rule-based AI
• Rules developed using only traditional machine learning techniques; No generative AI.

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.
• Logo Detection rules – These are used for advanced logo detection on documents and other marketing materials. 

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.
• 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:
  • Overpromissory language
  • Greenwashing language
  • Appeal to Minors language
  • General Advisory language
  • Spelling & Grammar