DALL-E 2, the image generator that took the internet by storm after gifting us masterpieces like ‘raccoon playing tennis’ and ‘medieval pizza festival,’ has become more widely available.
Catch-up: Launched by OpenAI this past April, DALL-E 2 uses AI trained on ~650 million web sources to skillfully edit images or generate them from a simple user prompt.
What happened: OpenAI is now fast-tracking the application waitlist so DALL-E 2 can hit one million users within a few weeks—with a new pricing plan to start bringing in revenue.
- Previously, anyone was free to join the waitlist for this sci-fi tech, but access was only granted to researchers, developers, artists, content creators, and journalists. Now the door is open for more average joes.
Those with access can now also use the AI-generated images in commercial projects (like artwork you can sell) instead of just for research or (non-monetizable) artistic expression.
Yes, but: This whole thing is creating a ton of legal uncertainty since the technology makes it easy to create (and then sell) content that might be copyrighted—as evidenced by how many DALL-E 2 users love putting Homer Simpson in strange situations.
- By (basically) making copyright infringement much easier, DALL-E 2 could start generating litigation as frequently as it generates sweet images.
- One IP expert told TechCrunch that rights-holders could potentially ask to collect licensing fees for images OpenAI uses to train their systems.
Why it matters: As OpenAI tries to ramp up access to its technology, there are still questions about the ethics of image generation technology that might be alarming in an internet world already overrun by so many fake things passing for real.