TikTok attributed a "moderator error" to the delay in removing a disturbing livestream by gossip blogger Perez Hilton that remained on the platform for approximately 15 minutes, according to a report from WIRED. The Tuesday night broadcast showed Hilton, whose real name is Mario Lavandeira Jr., appearing unclothed and covered in what resembled blood while holding what looked like a knife, prompting alarmed viewers to alert police. The incident has reignited questions about how livestreaming platforms handle content that depicts self-harm in real time.
TikTok spokesperson Jamie Favazza told WIRED that the platform's automated systems flagged the livestream within minutes of it starting, but a mistake by a human moderator caused a holdup in both taking down the video and banning the account. The company reached out to law enforcement shortly after the broadcast was detected, Favazza said. TikTok didn't answer additional questions about what exactly the moderator error involved or precisely how much time the mistake added to the stream's removal. Some social media users claimed the livestream stayed up for 30 minutes or longer, but Favazza denied this, saying it ran for roughly half that duration and that Hilton's follow-up streams were stopped within 90 seconds.
The Miami-Dade County Sheriff's Office got "multiple calls regarding an individual who was livestreaming acts of self-harm on social media" and sent a crisis response unit along with mental health professionals to Lavandeira's residence, according to detective Samantha Choon. A public information officer for the sheriff's office separately verified to WIRED that the person involved was Lavandeira. Dante Rusciolelli, Lavandeira's manager and CEO of Golden Artists Entertainment, told WIRED that the team wasn't in a position to discuss TikTok's internal policies or its contact with law enforcement, adding that their attention remained entirely on Hilton's health and well-being.
The platform's official content rules ban "content that shows, promotes, or provides instructions for suicide or self-harm" as well as graphic, violent, or disturbing material. TikTok has said it relies on a mix of automated systems and human reviewers to screen for such content. The company moved to a new ownership structure in January, with a recently created joint venture taking over responsibility for content moderation and data security in the United States. This isn't the first occasion that TikTok and livestreaming platforms more broadly have faced criticism for inadequate moderation of suicide or self-harm content. In 2020, The Intercept reported that a young Brazilian man had broadcast his own suicide on the platform, and later that year TikTok struggled to scrub graphic footage of Ronnie McNutt's final moments after he took his own life on Facebook Live.
The incident highlights the persistent challenge livestreaming platforms face in balancing real-time content with immediate safety intervention. Even when automated systems work as intended and flag problematic broadcasts quickly, human decision-making remains a critical—and potentially fallible—link in the moderation chain, particularly when seconds can mean the difference between early intervention and thousands of viewers witnessing distressing content. For platforms that prioritize live interaction as a core feature, the technical and human infrastructure required to respond within moments rather than minutes remains an unsolved operational challenge. The stakes are especially high when viewer reports and automated flags converge, creating pressure points where moderator judgment becomes the final gatekeeper between harmful content and mass audiences.