OpenAI brings DALL·E 3 to ChatGPT Plus

OpenAI just brought DALL-E 3 to ChatGPT Plus and Enterprise users, completely removing the hassle of wrestling with complex image prompts.

Now, generating an image is as simple as talking to ChatGPT. You describe what you want in plain text, ChatGPT creates visual options, and if you want tweaks, you just reply in the chat to refine the details.

Hands, text, and real conversation

DALL-E 3 fixes the biggest headache in AI image generation: listening comprehension.

By training a cutting-edge image captioner to re-label training images with richer descriptions, OpenAI made DALL-E 3 dramatically better at following long, intricate instructions. The new model reliably renders fine details that historically broke AI image tools, including hands, faces, and readable text, alongside native support for widescreen and portrait aspect ratios.

OpenAI also doubled down on guardrails. A multi-tiered safety stack scans prompts and resulting images to block violence, adult imagery, and hate speech.

To keep lawsuits at bay, DALL-E 3 automatically declines requests mimicking the style of living artists and restricts depictions of public figures. OpenAI also rolled out an opt-out program for creators who want their work excluded from future model training.

Alongside the release, OpenAI revealed an internal provenance classifier aimed at spotting DALL-E-generated visuals. In early tests, it hits over 99% accuracy on unedited images and stays above 95% even after resizing, cropping, JPEG compression, or slight text overlays. It is not infallible, but it is a major step toward practical AI provenance.

Why it matters

We think this is the moment generative art moves from a specialized hobby to an everyday utility.

Midjourney still wins fans for pure artistic styling, but requiring users to master prompt engineering and Discord commands feels outdated next to this. By letting ChatGPT translate casual user intent into detailed internal prompts, OpenAI eliminates the prompt barrier for everyday writers, marketers, and developers.

This completely rewires user expectations around editing. Instead of hitting reroll and praying, asking for changes in natural conversation makes visual asset creation feel like working with a junior designer.

However, the strict guardrails are a double-edged sword. OpenAI's hard refusal to emulate living artists and its aggressive safety filters reduce copyright liability, but they also limit creative range. For mainstream productivity, this is a massive leap; for artists looking for edge and rebellious aesthetics, the cage might feel a little tight.