there's a yeti in tech journalism.
His name is: Deepfake fraud.
It has been sighted several times, but its existence is unproven.
"Forbes" reported a few days ago about this Yeti: Criminals are said to have imitated the voice of a company director with the help of deepfake technology and thus got a bank manager on the phone to transfer 35 million dollars to a certain person.
The bank clerk in the United Arab Emirates thought he recognized the voice because he had spoken to the director before.
In addition, he had e-mails from both the director and the recipient of the money in his mailbox, from which it emerged how much money had to be transferred where for a supposedly pending company takeover.
The source of the article is a court document from the USA, the answer to a request for assistance from Dubai because part of the money had flowed into a US account.
There is only one sentence in this document about the alleged scam used by the perpetrators: "The Emirati investigations have shown that the defendants used deep voice technology to simulate the director's voice."
Deep Voice refers to deep learning, an application of so-called artificial intelligence (AI), which in this case is said to have learned to falsify a voice using training material.
In itself, that is possible nowadays, at least with predetermined sentences.
What exactly the artificially generated voice is supposed to say has to be said or - as I know it - typed in for text-to-speech output.
On the phone, this should happen in real time without the other person noticing, even if he asks questions.
The "Forbes" journalist, whom I appreciate very much for his articles, had tried to get more information from Dubai, but there was no answer.
Even so, his article was published.
In 2019 there was such a story about the electronically imitated voice of a company boss, it first went through German media, then ended up in the »Wall Street Journal« and is still part of deepfake folklore today. The case was made public by the insurance company of the company defrauded at the time. But she had no evidence for her deepfake theory, admitted a spokeswoman when I asked her about it.
Half a year ago a variant appeared in the Guardian, among others.
Accordingly, fraudsters had convinced several politicians in Europe via deepfake video that they were talking live via Zoom with Leonid Volkov, Alexei Navalny's chief of staff.
I questioned the story here in the newsletter back then because, despite all the technical advances, I still considered it unlikely that a real-time video deepfake would go unnoticed.
Shortly afterwards, it turned out that it was just a human doppelganger who had tricked the MPs.
As long as there is no corresponding evidence, I therefore assume a Yeti sighting in the case of the Emirates as well.
External links: three tips from other media
"Worse than Google: Which data are collected by alternative Android manufacturers" (4 reading minutes)
"Heise" has summarized a study according to which Android smartphones from well-known manufacturers permanently transmit data to third parties - even if users activate the strictest data protection settings.
Operating systems like / e / -OS show that there could be another way.
"How the 'pastiche' came into copyright law and what it means for creative work" (5 minutes' reading)
There is a new term in copyright law: pastiche.
Pa ... what?
iRights.info explains what it is and what is allowed with it: »Caricature and parody make ridiculous or exaggerate.
The pastiche, on the other hand, is a friendly, benevolent form of recognition and confirmation. "
»Bohemian Catsody« (video, English, 5:30 minutes)
When cats sing a song by Queen ... Sorry, but we haven't had any cat content in the newsletter for a long time.
I wish you a pleasant week!
Patrick Beuth