The world is an extraordinarily complex place. I feel like we’re only just getting used to the idea that images are doctored, and now AI deepfakes make it hard to trust. It’s important to listen to that voice in your head, as one Ferrari executive did, because the company likely avoided a costly and embarrassing mistake by outwitting the scammers in a delightfully ironic way.
Akio Toyoda can’t blame artificial intelligence for the loss of trust in his company. No, my friends, this is a classic case of a company doing too much, too fast. Japanese automakers as a group are generally in a strange situation, which is why there are rumors that Mitsubishi will join the Honda/Nissan software alliance.
And finally, let’s end this episode of The Morning Dump with some news on the upcoming Cadillac Celestiq.
How a book defeated an AI deepfake
The scene painted by this Bloomberg The article is quite disturbing. An anonymous Ferrari executive receives a series of WhatsApp messages from CEO Benedetto Vigna, telling him that a major deal is in the works and that the utmost secrecy is required.
It’s a bit risky, involves a lot of money and has to be kept secret for now, the report says. Vigna is using a different number because he doesn’t want anyone to know. The director should expect to sign a confidentiality agreement as soon as possible.
Obviously, no one would want a leader to send a large sum of money because of an email and a WhatsApp message in this year of Logano 2024, but a phone call followed from Vigna:
The voice imitating Vigna was convincing: a perfect imitation of the southern Italian accent.
Vigna’s deepfaker began explaining that he was calling from another cell phone number because he needed to discuss something confidential — a deal that could run into issues with China and required an unspecified currency hedging transaction to be completed.
The leader was shocked and began to get suspicious, the sources said. He began to perceive the slightest mechanical intonation which only increased his distrust.
“Sorry, Benedetto, but I have to identify you,” the executive said. He asked a question: what was the title of the book that Vigna had just recommended to him a few days earlier?
I have to say that this simple article triggered a wave of thoughts in my brain.
First of all, an excellent report from BloombergSecond, there are plenty of apps that can create a voice AI bot in about 11 minutes, and there are plenty of public audio files from a CEO to make it work.
Third, the world is an extremely complex place, filled with cybercriminals. In the last few weeks, we’ve seen CDK Global hand over potentially hundreds of millions of dollars to hackers. Even applying for a job at a parts store five years ago was risky!
Fourth, excellent work by the executive committee here for having the presence of mind to fight all of this.
Fifth, and finally, the irony of how this was solved is almost too perfect. Not only was an AI social engineering/deepfake attack repelled by a book, but the book itself is so obvious that you’d think they were making it up: Decalogue of complexity: acting, learning and adapting in the world’s incessant evolution
Here is the description (translated):
Why does complexity always increase? What is the complexity dilemma? What does it mean to dance with complex systems? What are the implications of the law of necessary variety? What is the link between paradox and metamorphosis? These are some of the questions that this Decalogue attempts to answer. In the vast existing corpus on complexity, ten topics treated in a synthetic manner can represent for the novice reader an opportunity to discover an exciting theme and for the expert reader an opportunity to reread, with a different thread, key themes of the subject.
The book seems to be specifically about the complexity of systems as they relate to business, and here we see an example of how complexity (in terms of communication and business structure) almost contributed to a serious loss of capital. It’s crazy.
Akio Toyoda: “I Won’t Be A Director Next Year” If I Continue Losing Support So Quickly
Former Toyoda CEO and current President Akio Toyoda has resisted an attempt to oust him from his top position due to all of Toyota’s problems with certification and its admittedly grueling work culture.
Now, by Reuterswe hear from Toyoda himself that if he fails to turn things around, he could be out next year:
Last month’s result marked the lowest support rate ever for a director in Toyota’s history, the 68-year-old grandson of the company’s founder said in an interview with the automaker’s own media outlet.
“If it continues at this rate, I won’t be able to be a director next year,” Toyoda said.
Its approval rating among foreign institutional investors was particularly low, at 34%. Ahead of the meeting, proxy advisors Institutional Shareholder Services (ISS) and Glass Lewis both criticized Toyota’s handling of certification test violations.
I think that given the way this case is going, we shouldn’t jump to the conclusion that Toyoda is done for. It’s hard to imagine that investors would dump Toyoda after he made them so much money and because of his family, but it’s a sign that Toyoda is at least pretending to take what happened seriously.
Mitsubishi-Nissan-Honda?
Japanese automakers are currently enjoying varying degrees of success as the companies have generally taken a more cautious approach to electrification, automated driving and software.
It works for now, but what about the future? The Rivian/VW deal is proof that software-driven automated driving is extremely complex, and asking each company to try to create its own version is extremely expensive.
So it’s no big surprise to learn that Mitsubishi, which married Nissan a few years ago, could join the previously announced Honda/Nissan merger.
This comes via Nikkeiwho was the first to report the news:
Mitsubishi will work with Honda and Nissan to finalize specific details of the alliance. The three companies intend to standardize the onboard software that controls the vehicle. Nissan and Honda are expected to jointly develop the core software and discuss its use in Mitsubishi vehicles.
It would be utter madness for Mitsubishi to try to do this alone.
Every Celestiq you see will be special
With all the problems GM’s current-generation Ultium electric platform had when it launched, I assumed the luxurious $340,000 Cadillac Celestiq was a vehicle we wouldn’t see anytime soon.
Our friend Mark Phelan from Detroit Free Press I went to Cadillac’s new Cadillac House in Vanderbilt, Michigan to tour the place where fancy people can design their luxury cars to meet their luxury standards:
Each Celestiq owner will sit down with GM designers to select the materials and look of their car’s interior. A selection of fabrics, leathers, woods and other elements are hidden behind wood panels in the open-air space of Cadillac House, a former cafeteria that has been restored and expanded to create consultation, tour and design work spaces for customers.
Many luxury car manufacturers offer this service, although it’s not as common for American luxury brands. Also check out this little excerpt about the origin of the center’s name:
The building’s name pays homage to Suzanne Vanderbilt, who became one of the auto industry’s first full-fledged designers when GM hired her in 1955 while she was working at Pratt Institute. She was part of a group dubbed the “Damsels of Design” for publicity purposes. Ever wonder how long it takes for an idea to go from innovation to condescension? Now you know. Vanderbilt was initially limited to giving cars a “feminine touch,” but she stuck with GM well into the 1970s, working on a number of concept and production cars, including some Cadillacs.
Henry Payne would never do it…
What I listen to while writing TMD
Childish Gambino, aka Donald Glover, aka Bando Stone, has a new song out, so enjoy “Lithonia” and all that it portends for his future. Especially the horror movie-esque twist at the end. Also, as a side note, I just learned that “This is America” started out as a Drake-bashing song, which makes me appreciate it even more (TH and I, as Torontonians and Houstonians, respectively, earned our Drake enmity the hard way).
The big question
What question could someone ask you that would indicate you are/are not an AI deepfake?