A regional court in Munich ordered the carmaker to reimburse a customer the bulk of the 112,000 euros ($115,000) she paid for her Model X crossover, because its Autopilot (AP) function could not navigate city streets properly, and phantom braking issues proved a “massive danger” to urban traffic safety. 

According to Der Spiegel, the court chose not to accept Tesla’s argument that AP was never designed for that environment. Instead it ruled that customers cannot be expected to manually switch the system on and off again constantly; it would be too much of a distraction from driving. 

While it may be tempting to think Tesla is now potentially liable should other customers come forward, German jurisprudence does not generally work like that. 

Unlike the American system of common law borrowed from the British, Germany uses civil law heavily influenced from Napoleonic reforms. That means it does not rely on legal precedent, but rather treats each case individually and on its own merits, detached from past rulings. 

Furthermore, damage claims against companies are generally capped in Germany and never reach U.S. levels that can sometimes pose existential risks. So there’s little reason to believe this could result in an avalanche of damage claims.

However, the case does highlight the often misleading way Tesla has marketed its self-driving features in the past, as well as its inability to deliver autonomous cars after customers spent thousands of dollars and in many cases waited years. 

The U.S. National Highway Traffic Safety Administration has moved a step closer to recalling Teslas equipped with such self-driving features last month after an examination of accident data indicated it was undermining driver supervision. 

The technology is even more rudimentary in European countries like Germany, where Teslas offer little appreciable advantage over competitor models and in some instances can be even less advanced.

‘Infamous last words’

Most rivals from the tech and auto industries have tempered their enthusiasm for autonomous driving as the challenges of mastering each and every edge case became clearer over time, but that does not include Musk.

For years he’s claimed Tesla is on the verge not only of delivering full self-driving (FSD), but doing so at a lower cost than its competitors by eschewing the use of expensive sensors like LiDAR in favor of industry-leading artificial intelligence.

Musk marketed his stock and his product on this promise, arguing three years ago it would be “financially insane to buy anything other than a Tesla,” because FSD would turn the car into an appreciating asset. 

Since then, however, he has admitted on various occasions that self-driving was more difficult than he anticipated after he was fooled by repeated false dawns (although Musk prefers to use a mathematical term—“local maximum”—to describe an initial burst of progress followed by rapidly diminishing returns). 

“You start getting to these what I call local maxima, where you don’t realize basically how dumb you were,” he told TED’s Chris Anderson in April.

Just weeks before he said that, his senior director of A.I. went on sabbatical—never to return, it turned out.

For the moment, Tesla is about to airdrop its v10.13 beta to those select customers, last numbered at 100,000 roughly, who have access thanks to high safety scores achieved under Tesla’s own proprietary ranking system.

The real problem for Musk could be if he ever faces a class action lawsuit in the U.S. by Tesla customers demanding compensation for a system that since January costs $12,000 but has yet to deliver on its claims and remains mired in controversy.

“I don’t want to blow your mind, but I’m not always right,” Musk went on to tell TED’s Anderson, when asked why he still cannot meet his 2017 target for a zero-intervention drive from Los Angeles to New York. “Admittedly these may be infamous last words, but I actually am confident that we will solve [full self-driving] this year.”