You might need to add NSFW filtering to the aigateway API

(edited)

Dear Vercel team, as far as I know, you generously provide a free monthly $5 credit for aigateway users. However, some SillyTavern players have noticed this and appear to be registering numerous accounts to obtain substantial AI API quotas (it seems your phone number verification only checks whether the number exists—anyway, they clearly have ways to register in bulk). Given that they often use the API to generate or translate NSFW content, you might consider implementing keyword filtering. Of course, you probably have smarter methods to prevent such abuse. I hope this information is helpful.

Vercel Staff

Thanks for reaching out about this. I shared your report with internal teams to investigate solutions to prevent this kind of abuse in the future

You’re welcome. As far as I know, they mainly use it for SillyTavern generation with English and Chinese or for translating Japanese NSFW works. Hope this addition is helpful to you.

reply to #4

Hello,

Although the use of Google and Claude models was previously halted due to free credit limitations, the SillyTavern community has now started using the free quota of the AI gateway again by registering in large numbers to translate and generate NSFW content. Perhaps because the scale of this abuse is relatively small, Vercel's technical team does not want to add filter words for this matter? I just want to make sure that Vercel is aware of this situation. I wish you all the best in your work.

Vercel Staff

Thanks for reaching out again about this. I should have posted an update.

The team is planning some solutions to help mitigate abuse of the AI Gateway. There is not a specific arrival date for that just yet.

This topic is closed. It no longer accepts new replies.