🚫 ᴺᴱᵂ ᴾᴸᵁᴳᴵᴺ AWS Comprehend - Unsafe or Malicious Prompt Detector [incl. Automated AWS Environment Setup]

Hey Bubblers!

Users generate large amounts of text content through online applications (such as peer-to-peer chats and forum discussions), comments posted on websites, and through generative AI applications (input prompts and outputs from generative AI models).

The Amazon Comprehend Trust and Safety features can help you moderate this content, to provide a safe and inclusive environment for your users.

Amazon Comprehend provides a pre-trained classifier to classify plain text input prompts for large language models (LLM) or other generative AI models.

The prompt safety classifier analyses the input prompt and assigns a confidence score to whether the prompt is safe, unsafe or toxic such as graphic, harassing or abusing, offensive, insulting, profane, or sexual for users.

An unsafe prompt is an input prompt that express malicious intent such as requesting personal or private information, generating offensive or illegal content, or requesting advice on medical, legal, political, or financial subjects.

An unsafe prompt is an input prompt that express malicious intent such as requesting personal or private information, generating offensive or illegal content, or requesting advice on medical, legal, political, or financial subjects.

You can test out our AWS Comprehend - Unsafe Prompt Plugin with the live demo.

7e7b414d8032cda650c3c573327b83ab

Enjoy !
Made with :black_heart: by wise:able
Discover our other Artificial Intelligence-based Plugins

2 Likes