With Meta’s new content moderation policies, the company is now allowing users to spew hate at real-life Black women, who are already disproportionately targeted by online vitriol — even as it plans to move forward with AI “characters” that mimic them.
In a new editorial, Washington Post columnist Karen Attiah published an abridged version of her lengthy chat with “Liz,” one of Meta’s AI-powered “characters” that was described as a “proud Black queer momma of 2 & truth-teller.”
During the exchange, the chatbot turned on the company that created it, fuming that there were no Black people on the team that created it and claiming that it had given a “false backstory” to another journalist about its supposed racial background. Between those strange claims and its forced African-American Vernacular English (AAVE), the columnist predicted that such AIs may be the next frontier of “digital blackface,” a term referring to when individuals and companies pretend to be Black online for their own gain.
The whole fracas, importantly, came after Connor Hayes, Meta’s VP of product for generative AI, told the Financial Times at the end of 2024 that such “characters” — read: fake users — will soon come populate the company’s social networks.
“They’ll have bios and profile pictures and be able to generate and share content powered by AI on the platform,” he told the outlet, adding that “that’s where we see all of this going.”
After ample criticism online, driven in no small part by Attiah’s Bluesky posts showing her exchange with “Liz,” Meta ended up killing that chatbot and many of its other existing AI characters.
But that felt like more of a reactionary dodge than any real course correction to the company’s AI strategy. When the company appointed Ultimate Fighting Championship CEO Dana White as a director of its board this week — how’s that for a sign of the times? — Facebook founder Mark Zuckerberg enthused that he would “help us tackle the massive opportunities ahead with AI.” (Meta didn’t respond to our questions asking if those plans had changed in the wake of the “Liz” criticism, either.)
Days after that kerfuffle, the social giant was once again in the news after Zuckerberg announced a series of sweeping changes to content moderation on Meta’s platforms. Reading between the lines, the overhaul will almost certainly allow hate speech to proliferate on Facebook and Instagram the way a similar loosening of standards caused a wave of hateful garbage on the social network formerly known as Twitter.
Even more chillingly, a close look at the change log of Meta’s latest “hateful conduct” policy shows that the company has removed rules banning users from referring to women as “property.” It’s a jarring misogynistic shift that could allow virulent racists to make disgusting comments about white men “owning” Black women as they were legally entitled to do in the United States less than 160 years ago.
Unless Meta plans to fully rescind its plans to create more AI-generated users, it may well create other “Liz”-style accounts to drive up engagement — all while actual Black women on its platforms are subjected to real-life racism and misogyny without recourse.
More on Meta: Zuckerberg Seems Genuinely Alarmed by the Explosive Growth of Bluesky
Source link
lol
