When an audio clip allegedly featuring a local school principal making derogatory remarks went viral, it led to death threats against the educator and caused turmoil in a suburb near Baltimore. However, it was soon revealed to be a fake, manipulated by AI. Yet many still believe it’s authentic.
The clip began with what sounded like a secret recording of Pikesville High School Principal Eric Eiswert expressing racist and antisemitic views. It gained traction online, particularly in Pikesville—a community with significant Black and Jewish populations. Following its release, Eiswert was placed on paid administrative leave pending an investigation.
Alfie Malone, a local resident, saw the clip circulating on social media and assumed its authenticity. “In the back of your mind, you think this is probably how people really feel about us,” he said, and he re-shared the clip after hearing it sounded like other recordings of the principal.
Despite its widespread sharing, many did not realize that the audio was AI-generated. The Baltimore Banner’s education reporter, Kristen Griffith, initially approached the story believing it was a straightforward case of a teacher’s misconduct. However, after reaching out to Eiswert’s union, she learned that he had not made the comments and suspected the audio was fake. When she published this information, it only fueled skepticism and backlash from the community.
As tensions escalated, local police launched an investigation. Pikesville High staff expressed concerns for their safety, fearing the school might be surveilled. Eiswert’s reputation suffered significantly, leading to increased security for him and the school.
In April, police confirmed that the recording was indeed fake and charged 31-year-old Dazhon Darien, the school’s athletics director, with various offenses, including theft and stalking. Investigators claimed Darien created the deepfake to discredit Eiswert amid performance issues and a pending investigation into his alleged theft from the school.
The authenticity of the audio was hard to debunk because it lacked visual cues typical of AI manipulation. The content included specific jargon and names that lent it credibility. Still, careful listeners could detect inconsistencies in the editing and a monotonous tone.
The main reason many believed the audio was real, according to Malone, was that it resonated with their lived experiences of racism. Even after the revelation that it was fake, the impact lingered in the community. Eiswert has since taken a new position elsewhere, but the fallout remains.
For residents like Sharon, the inflammatory nature of the comments made them fearful, even after learning the clip was AI-generated. The incident has highlighted broader concerns about the dangers of misinformation, particularly in communities sensitive to issues of racism and discrimination.
Despite social media platforms claiming to have policies to manage AI-generated content, such measures often come too late, after misinformation has spread widely. Malone noted that many still believe the clip’s authenticity, emphasizing the need for vigilance before sharing potentially harmful content. “I’ve been burnt by the fire once. I’m not touching the stove again,” he concluded.