Archive — US — techcrunch
ChatGPT’s image-generation feature gets an upgrade
OpenAI announced a significant upgrade to ChatGPT’s image-generation capabilities, leveraging the GPT-4o model, allowing for native image creation and modification. This new feature is available for Pro subscribers and will soon extend to Plus and free users. GPT-4o offers improved accuracy and detail compared to its predecessor, DALL-E 3. OpenAI emphasizes respect for artist rights in its training data usage.
Key points
- OpenAI CEO Sam Altman unveiled an upgrade to ChatGPT’s image capabilities on a livestream.
- ChatGPT can now create and modify images using the GPT-4o model.
- This upgrade is the first major change in over a year.
- GPT-4o improves image generation speed and detail compared to DALL-E 3.
- The feature is available for $200/month Pro plan users and will roll out to Plus and free users.
- OpenAI trained GPT-4o on proprietary and publicly available data while respecting artists' rights.
- An opt-out form is available for creators to remove their works from training data.
Why it matters
- The upgrade follows the launch of Google's Gemini 2.0 Flash, which encountered issues with copyright violations.
- There is an increasing industry focus on responsible AI development, particularly regarding intellectual property and training data.
- 2025 IT trends highlight the emphasis on AI ethics and data privacy in generative AI technologies.
The enhancement of ChatGPT’s image functionality marks a notable advancement in generative AI, prioritizing user capabilities while addressing ethical concerns regarding copyright and artist rights, reflecting broader trends in the tech industry.