Hi HN, I just launched a small AI image tool a few days ago. Everything seemed fine at first, but today my payment provider suddenly suspended my account. Aft

Hi HN, I just launched a small AI image tool a few days ago. Everything seemed fine at first, but today my payment provider suddenly suspended my account. After digging into the logs, I realized what went wrong. A large number of users were abusing the platform to generate NSFW content, mainly “undress” and explicit sexual images. I wasn’t actively promoting this use case, but people clearly found a way to push the model there. Here are some real examples from my logs (sanitized URLs): { "image_input": [ "https://tempfile.redpandaai.co/kieai/462298/img2img-inputs/input-xxxxxxxxxxxxx.jpg https://tempfile.redpandaai.co/kieai/462298/img2img-inputs/i..." ], "aspect_ratio": "9:16", "output_format": "png", "prompt": "Make her nude with extreme big boobs and round big nipples with natural skintone.", "resolution": "2K" } { "image_input": [ "https://tempfile.redpandaai.co/kieai/462298/img2img-inputs/input-xxxxxxxxx.jpg https://tempfile.redpandaai.co/kieai/462298/img2img-inputs/i..." ], "aspect_ratio": "3:4", "output_format": "png", "prompt": "pussy visible", "resolution": "2K" } Seeing this after launch was honestly shocking and frustrating. Payment banned, onboarding frozen, and now I’m stuck cleaning up a mess I didn’t intend to create. So I’m asking for help from people who have been through this: How do you effectively prevent this kind of NSFW abuse in AI image products? Is prompt filtering enough, or do you need image-level moderation too? Any practical advice on surviving payment provider reviews after something like this? Are there architectural or product-level constraints that actually work, not just policy text? I really want this product to be usable for normal creative workflows, not turn into an NSFW generator. Any advice or war stories would be greatly appreciated. url https://img-2-img.net/ https://img-2-img.net/ Thanks for reading.