Baltimore high school teacher accused of making racist AI deepfake of principal

Baltimore County Police Chief Robert McCullough speaks to reporters about the deepfake recording. Credit: AP

A US high school teacher has been accused of using artificial intelligence (AI) technology to create a fake racist and anti-Semitic audio recording of his boss, police say.

Dazhon Darien is alleged to have created the deepfake of Pikesville High School's principal Eric Eiswert's voice in response to a poor performance review and fears over his future employment.

The AI-generated clip, which featured fake audio stating that "ungrateful" black students "couldn't test their way out of a paper bag", went viral and was widely shared on social media.

Darien is now facing a number of charges, including theft, stalking and retaliating against a witness, according to court documents.

Police say the case is the first of its kind in the United States and have called for new laws to safeguard against the misuse of AI technology.

The development comes as experts continue to warn against the dangers that AI poses.

Police found that the clip had "profound repercussions", with Mr Eiswert being placed on leave and his family deemed to have been left at "significant risk".

The recording also triggered a wave of online trolling towards Mr Eiswert, while police officers were assigned to his home to provide security.

Darien was arrested as he attempted to board a flight at Baltimore/Washington International Thurgood Marshall Airport, police said.

The Baltimore County school system has recommended his immediate dismissal.

Pikesville High School, in Baltimore. Credit: AP

Why is the recording so concerning?

In recent years, cases of AI-generated disinformation have become more commonplace - many of which have been audio based.

Experts believe this is partly due to technology improving so quickly, but also because human ears cannot pick up signs of manipulation as easily as video or image discrepancies.

Some people have cloned the voices of purportedly kidnapped children over the phone to get ransom money from parents, experts say.

During this year's New Hampshire primary, AI-generated phone calls were used to impersonate US President Joe Biden's voice and dissuade Democrats from voting.

In 2023, meanwhile, fake audio of Labour leader Sir Keir Starmer and London Mayor Sadiq Khan was circulated during the Labour Party Conference.

Experts have repeatedly warned of a surge in AI-generated disinformation targeting elections in 2024.

Want a quick and expert briefing on the biggest news stories? Listen to our latest podcasts to find out What You Need To Know…

What can authorities do to prevent it happening again?

Most providers of AI voice-generating technology say they prohibit harmful usage of their tools, but self enforcement varies.

Some firms require a type of voice signature or ask users to recite a unique set of sentences before a voice can be cloned.

Facebook parent Meta and ChatGPT-maker OpenAI only allow a small group of trusted users to experiment with the technology because of the risks of abuse.

Some experts have called for companies to require users to submit phone numbers and credit cards so they can trace back files to those who misuse the technology.

Others have suggested a requirement for recordings and images to carry a digital watermark.

Another challenge is finding an international agreement on ethics and guidelines, according to Christian Mattmann, director of the Information Retrieval & Data Science group at the University of Southern California.

"People use AI differently depending on what country they're in," Mr Mattmann said.

"And it's not just the governments, it's the people. So culture matters."

Have you heard our podcast Talking Politics? Every week Tom, Robert and Anushka dig into the biggest issues dominating the political agenda…