Last updated: April 4, 2026
You type a prompt into ChatGPT expecting DALL-E to generate an image, and instead you get "Your request was rejected," "I wasn't able to generate this image," or "Content policy violation." Or the generation starts — you see the loading indicator — and then it just stops with a failure message and no image. In some cases the image generates but comes out as a solid color or a garbled mess.
DALL-E failures fall into two broad categories: content filter rejections (where your prompt triggered a safety filter) and technical failures (where the generation process itself broke down). Each category has different causes and different fixes. This guide covers both, explains what each error message actually means, and gives you concrete steps to get your images generating again.
"Your request was rejected" or "Content policy violation" → A word or phrase in your prompt triggered a content filter. See Fix 2 (rephrase prompt).
"I wasn't able to generate this image" → Technical failure on OpenAI's side. See Fix 1 (check status) and Fix 3 (clear browser).
Loading indicator spins then disappears with no image → Generation timed out. Check Fix 1 for ongoing incidents.
Image generates but is blank, black, or corrupted → Rendering error. Retry the same prompt; if it persists, see Fix 3.
"You've reached your limit" → Daily generation quota exhausted. Wait for reset or use the API. See FAQ.
ChatGPT responds normally but won't generate any images → Check your Plus subscription status. See Fix 4.
Why this matters: OpenAI maintains separate status indicators for their different services — ChatGPT, the API, and DALL-E image generation each have their own uptime tracking. It's entirely possible for ChatGPT's text functions to work normally while the image generation service is down. This is one of the most common causes of sudden DALL-E failures.
Go to status.openai.com in your browser. Look specifically for any incidents marked against "DALL-E" or "Image Generation" — these will be listed separately from the ChatGPT service status. If there's an active incident, no amount of prompt editing or browser clearing will help. Note the estimated resolution time and come back later. Most image generation outages resolve within an hour or two.
Why this matters: DALL-E's content filters are keyword and pattern based. They scan your prompt for specific words and phrases associated with prohibited content categories: graphic violence, sexual content, photorealistic images of real named individuals, content involving minors, and certain other sensitive topics. Even an innocent prompt can be rejected if it happens to contain a triggering word.
Here's how to diagnose and fix a filtered prompt:
Why this matters: ChatGPT's image generation interface is a web application, and corrupted browser cache can cause the generation process to fail at the display stage — meaning the image may actually generate on OpenAI's servers but your browser fails to receive or render it correctly. A cache clear forces a fresh connection.
Press Ctrl + Shift + Delete (or Cmd + Shift + Delete on Mac) to open browser clear data settings. Select "Cached images and files" and set the time range to at least the last hour. Clear the cache and reload chat.openai.com. Try your image generation prompt again in a new conversation thread.
Also try opening ChatGPT in an incognito/private window. If image generation works in incognito but not your regular browser, a browser extension (particularly ad blockers or content filters) may be interfering.
Why this matters: DALL-E 3 image generation within ChatGPT requires an active ChatGPT Plus subscription. If your subscription payment failed, your trial period ended, or there was an account issue that downgraded your access, you'll lose image generation capabilities — but the text chat features continue working, which makes the problem confusing.
If you're on a free plan, you have limited access to image generation as part of your overall message quota. Free users get fewer image generations per day than Plus subscribers.
Why this matters: Very long or highly complex prompts with many simultaneous requirements can sometimes cause generation failures — particularly when the prompt contains conflicting elements or describes scenes that are technically difficult to render coherently. DALL-E's processing has limits, and overly complex prompts occasionally time out or produce corrupted output.
If your prompt is more than three or four sentences long, try breaking it into its core elements. Start with a simple, short version of what you want and gradually add detail in subsequent generations. You can also ask ChatGPT to help you optimize your prompt — type "Help me write a better DALL-E prompt for: [your idea]" and use its suggestion.
Why this matters: DALL-E is accessible both through ChatGPT's web interface and directly via the OpenAI API. These are different infrastructure pathways. When ChatGPT's web frontend is having issues, the API often remains functional. If you have an OpenAI API key, you can call DALL-E 3 directly and bypass any frontend problems.
For non-developers, a simpler option is to try OpenAI's image generation through a third-party app or tool that uses the OpenAI API, such as the DALL-E feature in Bing Image Creator (which uses OpenAI models), or other OpenAI-integrated apps. These alternative frontends can bypass whatever is causing failures on the main ChatGPT interface.
Q: Why does DALL-E reject my prompt?
A: DALL-E's content filters are keyword and pattern triggered. They flag prompts containing language associated with graphic violence, sexual content, photorealistic likenesses of real individuals, and content involving minors. An innocent prompt can be rejected if it contains a coincidentally flagged word. To fix this, rephrase using synonyms, remove the triggering term, or describe your concept from an artistic or abstract angle rather than a literal one.
Q: How many images can I generate per day with DALL-E?
A: ChatGPT Plus subscribers can generate up to 50 images per day with DALL-E 3. Free users get limited image generation as part of their general message quota. Via the OpenAI API, there's no hard daily limit — you pay per image generated. When you hit your ChatGPT daily limit, you'll see a message with a reset time (limits reset at midnight UTC).
Q: Why are DALL-E images low resolution?
A: DALL-E 3 generates images at fixed resolutions: 1024x1024 (square), 1024x1792 (portrait), or 1792x1024 (landscape). These are built into the model and can't be changed through prompting. For higher resolution, download the DALL-E image and run it through an AI upscaler like Topaz Gigapixel, Adobe Firefly's upscale tool, or a free web-based AI upscaler — most can cleanly double or quadruple the resolution.
Q: Can DALL-E generate real people's faces?
A: DALL-E restricts generating photorealistic images of specific named real individuals — celebrities, politicians, and public figures. You can describe a fictional person with detailed attributes (age, hair color, ethnicity, style), and DALL-E will generate that person without issue. The restriction specifically targets prompts that name or identify real people and request photorealistic likenesses of them.
Q: What's the difference between DALL-E 2 and DALL-E 3?
A: DALL-E 3 is a major generational improvement. It understands prompts with far greater accuracy and nuance, generates readable text within images (DALL-E 2 struggled with this), produces more detailed and coherent compositions, and handles complex multi-element scenes better. DALL-E 3 is the default model in ChatGPT Plus. DALL-E 2 is still available via the OpenAI API at lower cost and can be useful for simpler or more abstract image needs.
If DALL-E is consistently failing despite trying all six fixes, OpenAI's support team can investigate account-specific issues. Visit the OpenAI help center at help.openai.com to submit a support request or search the community forums for others reporting the same error message.