OpenAI has announced that users of the free tier of ChatGPT can now generate up to two images per day using the DALL-E 3 model. DALL-E 3, which launched in September 2023, was initially available only to those subscribed to ChatGPT Plus.
A key feature of DALL-E 3 is its ability to expand on shorter prompts. When users give ChatGPT a brief request for an image, the chatbot elaborates on the prompt, helping DALL-E 3 create a more accurate image. OpenAI shared examples to demonstrate how this process works. The new image generation feature is gradually becoming available to users. Some, including those using the ChatGPT Android app and web client, have already accessed it. After creating two images, users are informed that they’ve reached their daily limit and are given the option to either upgrade to ChatGPT Plus or wait 24 hours to try again.
Additionally, OpenAI has been working on a method to identify text generated by ChatGPT. This involves using an invisible watermark to subtly mark AI-generated content, which could help address issues like academic dishonesty. However, OpenAI is cautious about releasing this tool publicly due to potential challenges. They are considering the risks, such as the possibility of bad actors finding ways around the watermark and the tool’s potential negative impact on non-English speakers. OpenAI’s spokesperson mentioned that while the watermarking method shows promise, the company is still researching alternatives to ensure it’s effective and fair.
