What’s next for NSFW detection in Discourse AI

[Saif (Saif Murtaza)]

If you’ve seen our previous announcement, then you are aware that we are shifting to prefer LLM preferred solutions as they provide a vastly superior experience. Hence, coming soon we’ll be decommissioning the NSFW module in favor of Discourse AI Post Classifier - Automation rules

This will be a beta experience so expect changes to occur to the feature.

Why are we doing this?

For the same reasons…

@Saif

Previously using the Toxicity module meant…

• You were stuck using a single pre-defined model
• No customization for your community-specific needs
• Confusing threshold metrics
• Subpar performance

LLMs have come a long way and can now provide a better performing and customizable experience.

Whats new?

@Saif

Discourse AI Post Classifier - Automation rule can be used to triage posts for Toxicity (amongst other ways) and enforce communities to specific code of conducts. This means…

• Multiple LLMs supported for different performance requirements
• Easy to define what and how content should be treated
• Customizable prompts for community-specific needs
• Flag content for review

and much more.

To assist with the transition we have already written out guides

• Setting up toxicity/code of conduct detection in your community
• Setting up spam detection in your community
• Setting up NSFW detection in your community

We will add another guide specifically for NSFW detection to help with the transition.

What happens to NSFW?

This announcement should be considered very early, until we are ready to decommission you can continue to use NSFW. When we do it, we will be decommissioning the module, and removing all code from the Discourse AI plugin and associated services from our servers. The deprecation will only happen after AI Triage is ready to handle images which will be done soon.

[Saif (Saif Murtaza)]

FYI about the following change

@Saif

Just following up here to note that AI triage no longer needs the following thanks to @Roman_Rizzi work on improving the prompts

@sam

It should contain a %%POST%% placeholder that will be replaced with the post content.

[Saif (Saif Murtaza)]

@Saif

AI Triage is ready to handle images which will be done soon.

This part is now complete, configure and use a vision-powered LLM via AI triage for Image detection

As a reminder, we have NOT yet deprecated the NSFW module. This will still take a bit of time

[Saif (Saif Murtaza)]

A new guide is available here

Setting up NSFW detection in your community

This is a how-to guide for setting up NSFW Images and text detection in your community using the Discourse AI Post Classifier. Required user level: Administrator Overview In this topic we are going to use Discourse AI Post Classifier to detect NSFW Images and text in your community. Using this guide, admins will be aware of such posts and take action accordingly. Note that the instructions here can be customized to your preference. Below is an example set…

[Saif (Saif Murtaza)]

As a heads-up, we are now hiding site settings for enabling/disabling Toxicity and NSFW. This is our ongoing effort as we continue to deprecate the features.

If you have these features turned on, it will still operate as normal. We have NOT fully deprecated the features yet.

If you have it turned off and would like to turn it on, you will now be unable to do so.