A missing sign costs Volkswagen 1.1 million euros. The Lower Saxony data protection authority imposed the penalty because the car manufacturer was caught in 2019 with a test vehicle that was filming traffic. The image data should improve the driver assistance system and prevent accidents. However, the necessary magnetic signs with a camera symbol were not hung on the car. In the future, however, an information symbol will not be enough. Cars are increasingly becoming rolling steady-cams, with cameras invisible to outsiders. They are indispensable for the assistance systems in order to train and improve them with huge amounts of video data. To do this, they have to be stored and transmitted: a red flag for consumer advocates who suspect violations of the European data protection regulation DSGVO. “We take a close look at what car manufacturers are doing,” says Heiko Dünkel, who heads the enforcement team at the Federal Association of Consumer Organizations.
A few weeks ago, the association filed a lawsuit against Tesla at the Berlin Regional Court. The “sentinel mode” of the electric cars constantly monitors the environment while parking and saves the data in certain cases. Although other manufacturers do not offer a function like “Sentry Mode”, their camera functions must also be checked while driving. “Tesla is just the beginning,” Dünkel told Handelsblatt.
According to experts, the lawsuit will not be heard in court until 2023. Either way, according to Thilo Weichert, former data protection officer for Schleswig-Holstein and member of the network data protection expertise, Tesla will lose Tesla. However, this will have fewer consequences for the automaker than for the owners of Tesla vehicles. When you buy, you take the risk. The Terms of Service states: It is “your sole responsibility” to ensure that “applicable laws and regulations” are observed. “This is a popular tool used by IT companies,” says Weichert, “to shift responsibility.”
Top jobs of the day
Find the best jobs now and be notified by email.
Tesla fully trusts the cameras
With all the sensors, a car can produce four to six terabytes of data per day. One terabyte of data requires 1,500 CDs to store the information. Most of the information is problem-free from a data protection point of view. The vehicle does not save the data. But those that are transmitted and stored are problematic for privacy advocates. Is it necessary to raise them? And are they anonymized? “The need for advice is increasing enormously,” says Alexander Duisberg, who, as a partner and technology expert at the law firm Bird & Bird, looks after numerous car companies.
That Tesla is being charged comes as no surprise. The US company is not only a pioneer in electric cars, but also pursues a unique approach to autonomous driving assistance systems. Tesla does without radar and lidar, wants to reach level 4 or 5 exclusively with camera data. This is the unit of measurement for driver assistance systems, with one as the lowest and five as the highest level of autonomous driving.
Vehicles from Mercedes and Co. in Germany can currently drive up to 60 kilometers per hour with level 3, Tesla comes to level 2. In order to achieve greater driving safety, Elon Musk’s company has to use artificial intelligence (AI) on all conceivable “edge cases Prepare: borderline cases in traffic, such as cyclists driving the wrong way or construction sites that change.
The problem is the “edge cases”
Today’s assistance systems already recognize almost all situations correctly. But with fully automated driving, much higher success rates are required, no mistakes can be made. To do this, assistance systems must also master “edge cases”. Tesla has been collecting video data from its customers in the USA for many years to train the AI on it. Hundreds of employees evaluate the images and label them so that the neural network can understand them.
That gets data protectors like Weichert from the network data protection expertise on the scene. Almost two years ago, he pointed out possible violations of the GDPR by Tesla in a report. The former data protection officer criticizes the “sentinel mode” in particular: there is no need to collect data, you can’t just film everyone who walks past a Tesla. The manufacturer has not reacted since then, as Weichert tells the Handelsblatt.
The transmission of the data to the USA is also still a problem, the data protection regulations there are not in line with the GDPR. “Even if Tesla stores the data on a server in Germany or Europe, that’s not enough,” says Weichert. Because the USA would have access to the data “without legal control” due to laws such as the “Patriot Act”.
Anonymizing the data is expensive
A way out for the car companies: anonymize the data. However, there is a risk that important information for the corporations will be lost. For example, where is the pedestrian looking? This can be decisive for the interpretation of the traffic, since the direction of view often allows conclusions to be drawn about the intention, says Steffen Heinrich, head of the video specialist Peregrine.
Heinrich used to work for VW and other car brands, founded Peregrine in Berlin in 2018. The start-up works for car suppliers and other customers. With its technology, it pixelates the data before it is stored. “License plates are easy to recognize and change,” says founder Heinrich. “It becomes more difficult with the faces of cyclists who are moving, for example, or with larger crowds – but that can also be solved.”
Another possibility is to alienate images or videos with synthetic data. In this way, an artificially generated face can be placed on the pedestrian, with the correct viewing direction. In this way, no important information is lost, and the neural networks can also be trained later with this data. But the alienation has disadvantages: “It’s more complex, needs more energy and computing capacity,” says Heinrich.
The best data protection is to collect as little data as possible. That is the approach of the Israeli start-up Autobrains, which is developing an AI for Continental or other customers that is supposed to get by with less data. The AI assigns so-called simple “signatures” to all objects: lines, colors or other structures in the image, without going into the last detail. “Recognition rates of 95 percent are not the challenge,” says Nils Berkemeyer, VP Corporate Development at Autobrains. “Our business is all about the last five percent, the unpredictable edge cases that you can’t train manually. This is where we are particularly strong.”
More: That’s why Tesla and Mercedes have so far failed in fully autonomous driving