Hello everyone,
First of all, thank you for your support — it's truly appreciated, especially during what have been challenging times for DDG. Many of you have noticed that our content moderation has become noticeably stricter recently, and you deserve an honest explanation of why that happened.
The short version: we didn't have much of a choice.
What happened
Some time ago, our payment provider flagged DDG for compliance issues, because some users were able to generate sexual content on the site. That kind of content has always been against our own rules — but it also violates our payment provider's policies, regardless of how it's tagged, where it appears, or whether it's public or private. The situation escalated, and we were ultimately hit with a very painful financial penalty.
More importantly, we were put on notice: either we significantly strengthen our content filters, or we risk losing our ability to process payments altogether. For a platform like ours, losing payment processing wouldn't just be an inconvenience — it would put us out of business.
This is the primary reason we've hardened our moderation system. It's not a change in our values or our vision for DDG — our stance on explicit content has been the same for years, as we explained in our earlier post Creativity Within Boundaries. What changed is that the cost of even occasional slip-throughs is now something we simply cannot absorb.
Why some filters aren't ours
There's a second, separate source of blocked prompts that we want to be transparent about. Some of the models we offer — most notably OpenAI's GPT Image 2 — come with their own built-in moderation filters, enforced on OpenAI's side. These filters frequently block prompts that include artists' names, due to copyright-related restrictions.
Unfortunately, there is very little we can do about those. When a prompt is rejected by the model provider itself, the decision happens outside of our systems. If a prompt with an artist's name keeps getting blocked, trying a different model on DDG or describing the style in your own words instead will often get you where you want to go. Our AI Art Assistant is great at this too — ask it to rewrite or reimagine your prompt, and it will capture the style you're after in words that pass the filters.
What this means in practice
Our filters now err on the side of caution. That means some perfectly innocent prompts and images will occasionally get blocked — false positives are the unavoidable price of a stricter system. We know how frustrating it is to have a harmless creation rejected, and we're continuously tuning the filters to reduce these cases without loosening the protections we're now required to maintain.
We ask for your understanding here. When something gets blocked that clearly shouldn't be, it's not censorship for its own sake, and it's not us second-guessing your creativity. It's a system that has to guarantee a hard boundary, learning to draw that line more precisely over time.
And if a harmless prompt of yours keeps getting rejected, send it to us at contact@deepdreamgenerator.com — we'll look into it. Real examples from you are the most useful input we have for tuning the filters.
Looking ahead
We understand that stricter moderation is not what anyone wants to hear, and we hope this post makes clear the position we're in. The reality is that we haven't been given many options: tighten our moderation, or lose the ability to keep DDG running at all. Between those two, we'll always choose the one that keeps this community alive.
DDG has been around for a long time because of you — your art, your feedback, and your patience. Thank you for sticking with us through this. We'll keep refining the system, and we'll keep being straight with you about the constraints we operate under.
Sincerely,
The DDG Team
Comments (126)