“Harry Potter” narrator Stephen Fry says AI was used to steal his voice

Actor Stephen Fry, the British voice behind the Harry Potter audiobook series, revealed that artificial intelligence was used to mimic his distinct accent for a documentary he knew nothing about.

“I didn’t say a word about it – it was a machine. Yes, it shocked me,” said Fry, who has narrated more than 100 hours of the popular JK Rowling series.

“They used my reading of the seven volumes of the Harry Potter books and from that data set an AI of my voice was created which created this new narrative.”

According to Fry, a snippet of the AI-generated audio was played at the CogX Festival in London last week, warning of the dangers posed by the technology Assets.

“What you heard was not the result of a mix-up. This is done through a flexible artificial voice where the words are modulated to match the meaning of each sentence,” said Fry.

“So I could read anything from a call to storm Parliament to hardcore porn, all without my knowledge and without my permission.” And what you just heard happened without my knowledge. When I heard about it, I sent it to my agents on both sides of the Atlantic and they were outraged – they had no idea such a thing was possible.”

The growing presence of AI has become a major theme in the ongoing writers’ strike that has paralyzed Hollywood.

Although Fry did not announce whether he would press charges against the unidentified companies behind the AI, he warned that the audio clip would show that “it won’t be long before full deepfake videos are just as convincing.”

“Tech is not a noun, it is a verb, it is always moving. What we have now is not what will be. When it comes to AI models, what we have now will advance faster than any technology we have ever seen,” he said.

“One thing we can all agree on is that it’s a really strange time to be alive.”

According to a recent study, deepfake technology – which swaps a person with another person’s likeness in an image or video – is already in use and deceiving people.

Last month, the University College London used a text-to-speech algorithm trained on two publicly available datasets to create 50 deepfake speech samples in English and Mandarin.

The samples were played to 529 study participants who tried to distinguish the real voices from the fake ones.

Participants were only able to detect fake speech 73% of the time – one in four cases – which only improved slightly after they received training in recognizing deepfake voices.

Additionally, X-rated celebrity deepfakes are circulating heavily online, despite Twitter/X’s policies against media and non-consensual nudity.

Most of these realistic videos are eventually removed from the platform, but not before they have garnered tens of millions of views.

A clip that received widespread attention on X in June showed 22-year-old influencer Addison Rae Easterling’s face on another woman’s body, lying seductively on a bed.

Before the account owner deleted the tweet after widespread outrage, it was viewed more than 21 million times and the thread contained additional explicit content from the deepfake.

Earlier this year, actresses Emma Roberts and Scarlett Johansson’s likeness was also used in a sexually suggestive ad for deepfake software, which was later removed by Facebook.

Meanwhile, a Texas teacher fell victim to digitally altered nude photos.

Deepfake content has also made its way into politics, forcing Google to require “all verified election advertisers” to “obviously disclose” if their ads use AI, amid the growing threat of misinformation from exploding technology.

Google’s decision coincides with warnings from other major AI companies that the technology could be used to undermine elections.

A deepfake image of Donald Trump resisting arrest and his wife Melania yelling at the police has already gone viral.

Another recently deployed deepfake was reported Trump hugs Dr. Anthony Fauci during the COVID-19 pandemic in an AI-produced campaign ad by Republican presidential candidate rival Ron DeSantis.

President Joe Biden has also been the target of several ads that used AI, including one from the Republican National Committee in April in which he celebrated with Vice President Kamala Harris after winning re-election to a second term.

The 30-second clip showed a happy Biden and Harris, then cut to harrowing scenes of China’s invasion of Taiwan, shuttered U.S. banks and crime-infested cities.