#96% of ai generated imaged are nsfw.

[aggressiveguitarnoises]

the more i find out about AI, the more im scared to fully turn to digital art as a career

[itmii-socialmedia]

Toxicity of the Social Media Scape: The Struggle to Regulate AI & Advertisements

Since the beginning of 2024, increasing amounts of AI content seem to be deepfake pornography and advertisements; 96% of deepfake videos in circulation are non consensual pornography.  This issue took center stage when explicit deepfakes of Taylor Swift began circulating on X & Meta Platforms; her name had to be restricted in search, but the images generated over 45 million views before actions were taken.  This matter extends past solely monitoring AI; Big Tech’s advertisement monitoring is severely lacking as explored by Mozilla’s testing of a myriad of platforms’ ad repositories in correspondence with Check First.

Table summating Mozilla's stress test results conducted on Major Tech's ad repositories. image via Mozilla

Mozilla’s report includes social media presences such as X, Meta, TikTok, Pinterest, and LinkedIn.  Platforms were evaluated on public availability, content of advertisement, targeting parameters, reached audience data, data accessibility, reliability, accuracy, and more.  Overall, all of the aforementioned companies failed to a certain degree to be considered fully functional.  Among the worst ad repositories were SnapChat and X, both lacking data and functionality; X was by far the worst platform.  Consequently, EU advocacy lead at Mozilla claims that these shortcomings “may be why the European Commission has included X’s ad repository in its formal proceedings against the platform under the DSA.”  In fact, the majority of platforms investigated were missing ads that were visible to users yet were not present in repositories and faced issues with public access to ad information.

AI girlfriend advertisement on Instagram Reels. image via Campaign

More recently,  Wired reports that Meta’s ad library hosts thousands of ads for NSFW AI companions being distributed across their platforms.  Despite Meta’s policies against “depictions of people in explicit or suggestive positions, or activities that are overly suggestive or sexually provocative”, the amount of ads of this nature is increasing.  Moreover, Wired found, using the same ad library, that at least 29,000 ads relating to AI girlfriends were published on Meta; Meta spokesperson Ryan Daniels claimed that “[they] continue to improve [their] systems, including how [they] detect ads and behavior that go against [Meta’s] policies.”  At the time of Daniel’s statement, Meta’s ad library had 3,000 ads for Ai girlfriends; 1,100 of which contained explicit content.  In connection with Mozilla’s ad repository test, there are most likely a myriad of advertisements for explicit AI girlfriends that are unaccounted for; Mozilla summarizes their stress test of Meta’s repository by stating that “we see progress since 2019 but are disappointed by accuracy errors and missing data fields in the web repository.” 

word count: 400