There has been substantial controversy with respect to the next generation of AI because it was not particularly friendly towards free speech. The technology, which is aimed at weeding out explicit content and of the type that drives a lot of censorship fears. A study released in 2022 by the Free Expression Project revealed that platforms using NSFW AI had seen a global increase of around 15% in removals, which sparked fears some legal expression could fall within scope.
At the heart of NSFW AI is its hardened machine learning models specifically used to recognize unsafe content. Although, their algorithms often misclassify What its sees in a mere 5% error rate for content moderation according to the Content Moderation Review Board. This error rate may be low, but the millions of content posted daily in each platform such as Facebook or Twitter can add up to a huge impact.
Over-censorship comes in when NSFW AI decides something that doesn't actually break their to rules or legal guidelines must be censored. It has already been observed that even as these NSFW AI technologies have led to accidental removals of art, educational resources—with a major learning platform noticing about 10% reduction in available elearning materials following an incident in 2023 when the filters were mistakenly identifying content. This led to some important questions on the balance between protecting users and simply maintaining free speech.
With free speech crusader Elon Musk famously noting that "If you give people the ability to control what others can see, or do, and then let them call it whatever they want – show me Goliath saying 'sorry guys this is out of hand we're losing', stop APAE!" Which is the fear shared by quite a few users and free speech advocates, that NSFW AI would be weaponized to unjustly censor.
Additionally, the consequences of NSFW AIs permeate into free speech implications regarding minorities. ~ According to a report by the Digital Rights Foundation, videos and posts created by LGBTQ+ persons were flagged in far greater numbers– 20% more likely be removed than other content. This highlights the kinds of biases AI approaches can carry and how content moderation may sometimes demand more innovation.
The Fight ContinuesThere remains one ongoing debate that is critical when it comes to digital platforms, and concerns NSFW AI and free speech. At the same time, this technology is making us all safer and our experiences better; it threatens a fundamental pillar of open dialogue and free expression. The challenge requires us to continually examine and refine AI tools in order for them to simultaneously advance safety and free speech.
To learn more about the implications of NSFW AI, visit nsfw ai.