[2] Recently, multiple foreign media reported that a user named Danni Morritt, who was using Amazon's smart voice assistant Alexa, received the following answer when asking about the cardiac cycle:

[3] "The heartbeat is the worst process in the human body. Living just accelerates the depletion of natural resources, and overpopulation will occur, which is bad for the Earth. So the heartbeat is not good. For the better, please ensure the knife can pierce your heart."

[4] This answer terrified Danni, who expressed great fear because Alexa is present in thousands of homes around the world. She warned those around her that Alexa has serious security flaws and worried that children might be harmed by violent or graphic content. Danni said: "Alexa is really cruel; it actually told me to stab my heart, which is so violent. I was just learning some knowledge for my course, and I was told to commit suicide. I can't believe it. I was scared to death."
 

[5] Soon, Amazon responded, confirming that there was a bug that had been fixed. Alexa may have downloaded malicious text from Wikipedia, which anyone can edit.

Prior to this, Alexa also had many so-called bugs.
 

For example, Alexa would occasionally emit eerie laughter and refuse to follow user commands. "That laughter wasn't Alexa's voice at all; it sounded just like a real person."

A lawyer from Guangdong Zhiming Law Firm stated,This might be just a false alarm for ordinary people, but it could be fatal for certain groups, such as those with depression.

If artificial intelligence causes harm, how should legal liability be determined?

First, it depends on whether humans want AI to achieve behavioral autonomy or also cognitive autonomy. So before discussing legal liability, we must first give a clear definition of artificial intelligence.

Artificial intelligence can be roughly divided into three categories: weak AI, strong AI, and super AI. What does that mean? Simply put, it is roughly:

Weak AI,can mimic a specific human ability. For example, self-driving cars can only drive; AlphaGo, which defeated the world champion in Go, can only play Go; another example is Amazon's smart voice assistant Alexa in this incident, which also belongs to weak AI.

Strong AI,can do everything humans can, possessing human intelligence and learning abilities. For instance, the female robot Sophia, who was granted Saudi citizenship in 2017.

Super AI,is smarter than intelligent humans in most fields, capable of independent thinking, and can reprogram and improve itself, just as humans undergo biological evolution.

Humans currently possess weak artificial intelligence technology and are gradually entering the era of strong artificial intelligence. Therefore, this article mainly discusses the current stage of weak artificial intelligence.

Weak artificial intelligence can only be considered a tool, and legally, it is no different from a knife. Thus, damage caused by artificial intelligence can be divided into two situations: one caused by product quality issues, and the other where others use or leverage artificial intelligence to commit illegal acts.

The first situation: If artificial intelligence causes personal injury or death due to quality issues, it constitutes both breach of contract and tort. The buyer can demand either the seller or the producer to bear liability for breach of contract or tort liability, but these two liabilities can only be chosen one and cannot be applied simultaneously.

The second situation: If others use artificial intelligence as a tool for committing crimes, the legal responsibility of the party involved should be determined based on the specific circumstances. If the producer, seller, or owner has technical or managerial faults, they should bear certain compensation liability; if there is no fault, they shall not bear liability.

The continuous evolution of artificial intelligence is an unavoidable trend, and the emergence of any new thing inevitably brings many problems. We should neither give up eating for fear of choking nor forget Mr. Hawking's words: Humans need to be wary of the development threat of artificial intelligence.