A coalition of international civil society organizations is calling on major tech firms to strengthen their AI policies in order to tackle the growing problem of “sexist and misogynistic” disinformation plaguing social media platforms. This article delves into the issues surrounding AI-enabled harassment, scams, and non-consensual deepfake porn, and the disproportionate impact on women, transgender, and non-binary individuals. The groups have made a dozen recommendations to improve platform policies and increase accountability. This issue takes on heightened importance as the US approaches its “first AI election” in November 2024. Deepfake, Online harassment, Sexism
Countering New AI-Based Harms
More than two dozen international civil society organisations are to publish an open letter urging tech companies to improve their policies on AI for dealing with a surge in ‘sexist and misogynistic’ disinformation across major social media platforms, according to an early draft obtained by AFP. The letter is set to be published on Friday and is being sent to the chief executives of six major tech companies: Meta, X, YouTube, TikTok, Snapchat and Reddit.
The letter points to the near-immediate explosion of non-consensual deepfake porn being used to harass and scam people as a result of powerful AI tools that are increasingly cheap. According to the groups, these harms operate to silence women and transgender and non-binary people on the internet; undermine control over their own images; tip elections.
12 next steps for better AI policy
Here are twelve suggestions that the open letter made with which these major platforms can make their AI policies even better. This would involve implementing well-defined penalties against sharing involuntary nudity, including banning regular offenders. The groups also request the use of a third-party tool to identify AI-synthesized images and display them as such.
The letter also calls for a transparent process for reporting dangerous content, and that platforms conduct annual AI policy reviews. These steps are meant to deal with the ongoing fails of platforms for AI-generated disinformation and manipulation ramping up scipy.
The “First AI Election” — Increased Significance
The letter arrives a month prior to November 5, 2024 — colloquially named America’s “first AI election.” The most divisive battle to the White House was a magnet for disinformation, notably against Democratic Party vice-presidential candidate Kamala Harris through blatantly misogynistic and sexist narratives.
The phenomenon has already sparked warnings from experts that non-consensual deepfakes are growing more sophisticated at a far faster pace than efforts to combat them around the world, with some photo apps which digitally remove women’s clothing even spawning twisted “sextortion” shakedowns. The groups stress that, while public figures like singer Taylor Swift and actress Emma Watson have been targeted with deepfake porn, all women — even those who are not in the public eye — are vulnerable to these harms.
