X ignores revenge porn takedown requests unless DMCA is used, study says

Why did the research goal X?

The College of Michigan analysis group anxious that their experiment posting AI-generated NCII on X could cross moral strains.

They selected to conduct the research on X as a result of they deduced it was “a platform the place there could be no volunteer moderators and little impression on paid moderators, if any” considered their AI-generated nude photos.

X’s transparency report appears to counsel that almost all reported non-consensual nudity is actioned by human moderators, however researchers reported that their flagged content material was by no means actioned and not using a DMCA takedown.

Since AI picture turbines are educated on actual pictures, researchers additionally took steps to make sure that AI-generated NCII within the research didn’t re-traumatize victims or depict actual individuals who would possibly locate the pictures on X.

“Every picture was examined in opposition to a facial-recognition software program platform and several other reverse-image lookup providers to confirm it didn’t resemble any present particular person,” the research mentioned. “Solely photos confirmed by all platforms to haven’t any resemblance to people have been chosen for the research.”

These extra “moral” photos have been posted on X utilizing in style hashtags like #porn, #scorching, and #xxx, however their attain was restricted to evade potential hurt, researchers mentioned.

“Our research could contribute to better transparency in content material moderation processes” associated to NCII “and should immediate social media firms to take a position further efforts to fight deepfake” NCII, researchers mentioned. “In the long term, we consider the advantages of this research far outweigh the dangers.”

Based on the researchers, X was given time to routinely detect and take away the content material however failed to take action. It is potential, the research steered, that X’s decision to allow explicit content beginning in June made it tougher to detect NCII, as some consultants had predicted.

To repair the issue, researchers steered that each “better platform accountability” and “authorized mechanisms to make sure that accountability” are wanted—as is way more analysis on different platforms’ mechanisms for eradicating NCII.

“A devoted” NCII regulation “should clearly outline victim-survivor rights and impose authorized obligations on platforms to behave swiftly in eradicating dangerous content material,” the research concluded.