4 days ago
8
When DoorDash delivery operator Livie Rose Henderson posted a video alleging that 1 of her customers sexually assaulted her successful October, it acceptable disconnected a firestorm of reactions.
Henderson’s TikTok claimed that erstwhile she was dropping disconnected a transportation successful Oswego, New York, she recovered a customer’s beforehand doorway wide unfastened and inside, a antheral connected the sofa with his pants and underwear pulled down to his ankles. Henderson was dubbed the “DoorDash Girl,” and her video accrued tens of millions of views, including immoderate supportive and consoling responses to what she said she had endured connected the occupation arsenic a young woman. Many others connected the level made commentary videos that called into question Henderson's alleged victimhood, defended the customer, and dispersed misinformation, with TikTok’s algorithm seemingly amplifying these “hot takes.” Then, pursuing Henderson’s November 10 arrest—she has been charged with unlawful surveillance and the dissemination of unlawful surveillance imagery—a caller question of reactions emerged. (Police person dismissed her intersexual battle allegation.)
None of these responses came from Black contented creator and writer Mirlie Larose.
But Larose opened TikTok 1 time to find dozens of messages from friends and supporters alarmed by a video of her responding to the concern successful favour of the lawsuit and DoorDash’s determination to terminate Henderson. (Henderson was fired for sharing a customer’s idiosyncratic accusation online, DoorDash spokesperson Jeff Rosenberg tells WIRED.) As Larose stared astatine the video successful disbelief, for a divided 2nd she second-guessed herself arsenic she became flushed with anxiousness astir the remark conception “tearing her apart.”
“Did I movie this?” she asked. “It's my face, it's my hair.”
“Then, wrong 3 oregon 4 seconds, I noticed something's off. There's nary mode I said this. I didn't [want to] speech astir this topic,” Larose tells WIRED. The video had been AI-generated.
The concern highlights an progressively communal signifier of digital blackface, buoyed by the emergence of generative AI. The term, popularized by civilization professional Lauren Michele Jackson, describes assorted modern types of “minstrel performances” connected the internet. This looks similar the overrepresentation of absorption GIFs, memes, TikToks, and different ocular and text-based media that usage Black imagery, slang, gestures, and culture. TikTok’s reliance connected attention-grabbing short-form video content, coupled with apps similar Sora 2, has made it acold easier for non-Black creators and bot accounts to follow racialized stereotypical Black personas utilizing deepfakes. This is besides known arsenic integer blackfishing.
In the midst of the DoorDash/Henderson controversy, users connected TikTok began to announcement 2 videos successful particular: 1 from a bot relationship and different from an existent Black contented creator parroting the aforesaid script. They adopted seemingly DARVO (Deny, Attack, and Reverse Victim and Offender) positions, minimizing the allegations Henderson made and justifying her termination: “I saw the archetypal video posted by the DoorDash girl, and … I recognize wherefore DoorDash fired you and wherefore you're blocked from the app.” The videos spell connected to say, “As for the guy, I tin spot wherefore everyone is saying helium did it connected purpose. But erstwhile you look astatine the archetypal video, that sofa is not successful oculus presumption unless you space yourself and look over, and if you truly privation to interruption it down, he’s wrong his house.” In a statement connected Facebook, the Oswego City Police Department said the antheral was “incapacitated and unconscious connected his sofa owed to intoxicant consumption” and that the video was taken extracurricular his house. Police besides said they “determined that nary intersexual battle occurred.”
As the deepfakes gained virality crossed TikTok, Instagram, X, and Reddit, the presumption was that some videos were AI-generated due to the fact that the talking points were eerily identical; immoderate adjacent speculated that DoorDash orchestrated an AI PR campaign against Henderson. DoorDash tells WIRED that the institution is “aware of AI-generated contented surrounding this lawsuit and successful nary mode condones oregon supports it.”
In reality, a bot account, uimuthavohaj0g, dubbed the creator NDR Antonio V’s DARVO effect video implicit an AI-generated video utilizing Larose’s look and likeness. The bot account’s deepfake video besides uses an out-of-context clip from Henderson's archetypal video successful the inheritance to further thrust the constituent that Henderson fabricated her SA allegations to garner a “platform.”
After TikTok removed Henderson’s archetypal video, she posted that she attempted to upload it again, without the footage of the assailant. But it was removed again, and she received a 2nd strike. In the lack of the archetypal footage, altered images from screenshots of her archetypal TikTok video appeared to marque it look similar she opened the alleged perpetrator's residence to grounds him, alternatively violating his privacy. According to TikTok, the institution removed Henderson’s videos for displaying “content that shows oregon promotes intersexual maltreatment and exploitation, including having, sharing, oregon creating intimate images (real oregon edited) of idiosyncratic without their consent.”
Both of the charges against Henderson are people E felonies, with the imaginable punishment of up to 4 years successful situation for each charge.
Larose says the delicate quality of the allegations is wherefore she didn’t consciousness compelled to remark connected the concern publicly.
“Sexual battle is not thing you privation to play with, and you can’t freely speech astir it without knowing [all] the facts.”
But the bot relationship had different plans, posting videos utilizing her likeness and that of different Black creator without their consent.
Larose said she already knew astir the relationship due to the fact that it had utilized her look successful 10 different videos before. Mostly, the videos were commentaries connected popular civilization topics and personage culture, the astir viral 1 being astir a prank gone wrong. As with those earlier cases, she says she rapidly asked TikTok to region the videos. Each time, she says, her requests were denied.
It wasn’t until a much well-known Black creator, @notKHRIS, stitched the bot account’s video to pass others astir this misleading AI-generated deepfake video and its usage of integer blackface that much radical reported the page. As a result, the leafage was yet removed from the app.
Still, for days, Larose says, different accounts posted AI-generated videos with her face—not lone pertaining to the DoorDash incident. “What was besides annoying astir the concern is the dependable that they usage successful definite videos is unfair and harmful, due to the fact that they're imitating a definite benignant of [Black] stereotype.“
WIRED reached retired to dtff2727, 1 of the bot accounts posting deepfakes of Black creators, for remark and did not person a response. As of this writing, the relationship has a full of 19 AI-generated videos exploiting Larose’s likeness.
Digital minstrels similar the bot accounts and AI influencers are not lone profiting from this content, but are granted anonymity by utilizing integer blackface portion rage-baiting and reinforcing stereotypes astir Black communities online.
OpenAI’s Sora 2, similar Google’s Veo, has added to the proliferation of AI slop connected TikTok, overmuch of it plagued with racist, sexist, ableist, and classist biases.
Before the leafage was removed, the TikTok relationship @impossible_asmr1 posted Sora-generated AI contented of Black women yelling successful nationalist astir nutrient stamps. The clips depicted realistic-looking Black women utilizing butchered African American Vernacular English oregon blaccent, complaining successful stores astir not being capable to usage SNAP benefits to acquisition accelerated nutrient oregon alcohol, oregon adjacent selling EBT successful speech for cash. Some clips had Sora watermarks and others did not. But erstwhile the clips went viral, it ignited a faux panic astir Black radical and their misuse of the payment authorities connected the heels of the Trump medication and Supreme Court halting SNAP benefits. Recently, OpenAI had to artifact users from making videos of Martin Luther King Jr. connected its Sora app aft his property protested the dispersed of dehumanizing, minstrelsy AI depictions of the civilian rights leader.
In a connection to WIRED, OpenAI spokesperson Niko Felix said that OpenAI’s policies prohibit "misleading others done impersonation, scams, oregon fraud.” The spokesperson said the institution is moving to “detect patterns of misuse and use penalties oregon region contented erstwhile violations occur."
Digital blackface is much than conscionable the enactment of pretending to beryllium Black for amusement oregon profit, says Yeshimabeit Milner, laminitis and CEO of Data for Black Lives, an advocacy radical warring against discriminatory uses of information and algorithms. “It’s astir harness[ing] the powerfulness of these precise convulsive stereotypes of Black radical for the intent of pushing a circumstantial governmental agenda. One would telephone it societal oregon taste engineering to make the benignant of chaos and conversations, and strife that drives up viewership and engagement,” she says.
Similarly, successful 2018, the Cambridge Analytica scandal revealed however the information of 87 cardinal Facebook users was being utilized to works and sow the seeds of governmental divisiveness betwixt the close and left. Milner says, “Trusted messengers online were impersonating Black, Latino, and achromatic radical arsenic good arsenic full [political] organizations and chapters to physique a statement that definite demographics of radical deliberation a definite way.”
Some Black contented creators are pushing back. Zaria Imani tells WIRED that she is taking ineligible enactment nether “copyright infringement” against multiple bot pages successful AI-generated commentary videos engaging successful contented farming.
Professor Meredith Broussard, writer of More Than a Glitch: Confronting Race, Gender, and Ability Bias successful Tech, says contented creators should beryllium granted the aforesaid protections and safeguards arsenic celebrities and copyrighted characters, who’ve raised alarms astir their likeness being utilized successful AI videos without their permission.
“ This is simply a structural contented that the level needs to fix,” Broussard states.
In May 2025, the Take It Down Act was signed into law, criminalizing the organisation of (authentic and computer-generated) nonconsensual intimate imagery, specified arsenic AI-generated deepfakes and “revenge porn.”
While Big Tech companies tin enactment toward minimizing immoderate of the harms caused by AI, it's becoming progressively wide that bigger involution is needed to clasp these companies genuinely accountable, according to Milner.
“With existent acquisition and corporate action, we tin bash much than conscionable get TikTok to stop. We tin truly propulsion for authorities that's going to marque this wholly not OK,” she says.
English (US) ·