3 ms·
DALL-E 2 was trained on approximately 650 million COPYRIGHTED image-text pairs SCRAPED FROM THE INTERNET, according to the paper that OpenAI posted to ArXiv. h
by NicoleJO 4y ago
DALL-E 2 was trained on approximately 650 million COPYRIGHTED image-text pairs SCRAPED FROM THE INTERNET, according to the paper that OpenAI posted to ArXiv. https://cdn.openai.com/papers/dall-e-2.pdf https://cdn.openai.com/papers/dall-e-2.pdf
That includes YOURS.
Have you been paid for it?