"""
Unfortunately, I can't generate images of people. My purpose is to be helpful and harmless, and creating realistic images of humans can be misused in ways that are harmful. This is a safety policy that helps prevent the generation of deepfakes, non-consensual imagery, and other problematic content.
If you'd like to try a different image prompt, I can help you create images of a wide range of other subjects, such as animals, landscapes, objects, or abstract concepts.
"""
It's unfortunate they can't just explain the real reason they don't want to generate the image:
"Unfortunately I'm not able to generate images that might cause bad PR for Alphabet(tm) or subsidiaries. Is there anything else I can generate for you?"
The rejection message doesn’t seem to be accurate. I tried “happy person” as a prompt in AI Studio and it generated a happy human without any complaints.
It’s possible that they relaxed the safety filtering to allow humans but forgot to update the error message.
when giving more context it replied:
""" Unfortunately, I can't generate images of people. My purpose is to be helpful and harmless, and creating realistic images of humans can be misused in ways that are harmful. This is a safety policy that helps prevent the generation of deepfakes, non-consensual imagery, and other problematic content.
If you'd like to try a different image prompt, I can help you create images of a wide range of other subjects, such as animals, landscapes, objects, or abstract concepts. """
What a weird rejection. You have to scroll pretty far in the article to see an example output that doesn't have a realistic depiction of a person.
It's unfortunate they can't just explain the real reason they don't want to generate the image:
"Unfortunately I'm not able to generate images that might cause bad PR for Alphabet(tm) or subsidiaries. Is there anything else I can generate for you?"
1 reply →
The rejection message doesn’t seem to be accurate. I tried “happy person” as a prompt in AI Studio and it generated a happy human without any complaints.
It’s possible that they relaxed the safety filtering to allow humans but forgot to update the error message.
1 reply →