The bot has some major issues. A lot of the things it’s flagging as nsfw are not actually nsfw.

Quick question, what does the user see if their image gets flagged by this bot?

They don’t see anything.

You don’t know until it’s posted.

... That's a huge problem imo...

Exactly.

The way the bot is currently set up:

Anything with a cig gets marked

Anything with a gun (presumably) gets marked

Anything with curse words like “damn” or worse (even if only once) gets marked

Obviously porn gets marked (hopefully, not willing to test this out but I think it will given what was just submitted to msmg)

Anything that contains violent words (“suicide” and “kill””) gets marked.

Anything mildly disrespectful gets marked if the bot considers it so.

However:

Spam does not and shouldn’t get marked.

Wrong stream (as far as I know) doesn’t and shouldn’t get marked.

Political memes: depends on the topic. Sometimes just mentioning the word is enough to get marked, other times you can skirt around it.

I’ve already had the bot mark criticism with no curse words.

Is this modbot?

No.

No, it’s an nsfw bot. It determines if your post is nsfw or not by testing certain values, half of which don’t even fall under the definition of nsfw.