When it comes to the use of AI models in everyday use-cases, there is a bit of a debate about who to blame in case there is an accident of some sort. For example, Tesla has repeatedly emphasized in the case of road accidents related to its self-driving cars in the past that the capability it offers in some of its cars is an assistive technology that is meant to aid the driver, not replace them. However, it now seems like the European Union (EU) is making it easier to sue AI model-makers for damages.
Reuters reports that the European Commission (EC) will unveil the AI Liability Directive on Wednesday to address grounds for liability when it comes to AI models and drones. In essence, this directive will make it easier for the victim to prove harm to their life, health, or property while dealing with or being affected by these technologies. This also applies to people being discriminated illegally during the recruitment process.
Basically, the burden of proof will be reduced on the plaintiff through a process called "presumption of causality". In this, the plaintiff will be required to prove that they were harmed by a product powered by these technologies due the manufacturer or product user's failure to comply with certain requirements. If this claim is then linked to an AI technology, the plaintiff could seek compensation in court.
Moreover, the plaintiff will also be able to demand "right of access to evidence" through which companies will be required to provide information about "high-risk AI systems" and identify the liable personnel.
This directive will not only apply to negligence from an AI product's user, but also to the manufacturer if they release software updates that make the product unsafe or if they do not fix cybersecurity issues in a timely manner.
Although the directive sounds quite strict for the defendant on paper, it will only become law if it is approved by EU countries and lawmakers.