European Commission proposes AI Liability Directive

What impact will the AI Liability Directive have on businesses using AI systems?

The question

What impact will the AI Liability Directive have on businesses using AI systems?

The key takeaway

The EU has proposed the AI Liability Directive which creates new rules for establishing liability in claims arising out of the operation of an AI system. 

The background

Existing legal frameworks around product liability were not designed with AI in mind. Claims do arise for damage caused by AI-based products such as drones, smart devices, robots and automated vehicles, but there have been challenges in assessing liability in those claims. Existing liability rules do not apply seamlessly to AI which makes it difficult for businesses to insure themselves against associated risk. This also hinders the ability for injured parties to claim legal compensation following damage, which in turn makes it difficult for consumers to trust and engage with new technologies. 
On 28 September 2022 the European Commission published a new proposal to tackle these problems by expanding existing product liability rules to address digital products, including AI.

The development

The AI Liability Directive adapts non-contractual civil liability rules in order to address liability arising from “wrongful behaviour” by AI systems. There are two key features of the AI Liability Directive. 

Firstly, national courts are provided with the right to order the provider of high-risk AI products (ie those which have an impact on safety or fundamental rights) to disclose relevant and necessary evidence about their product.

Second, the AI Liability Directive establishes a rebuttable presumption of causality between the damage suffered and the faulty output produced by an AI system. To take advantage of this, the claimant must show: (i) that the defendant (who is responsible for the AI system) is at fault by not complying with their duty of care; (ii) it is reasonably likely that the fault influenced the output of the AI system; and (iii) the output caused the claimant damage. 
The AI Liability Directive, therefore, is primarily concerned with fault-based liability. It should be considered alongside the Product Liability Directive (discussed in a separate Snapshot) which, amongst other things, adapts existing product liability rules to address “no fault” or “strict” liability for defective AI. 

The European Parliament and Council of Europe will need to agree the AI Liability Directive after which there will most likely be a two-year implementation period.

Why is this important?

The AI Liability Directive will impact businesses who operate in the EU and provide AI systems to the EU marketplace. There will be new avenues to claim compensation following harm caused by AI systems. In addition, the increased disclosure obligations may have significant implications for the protection of IP and trade secrets in innovative tech. Businesses must also comply with requirements connected to technical documentation, testing and compliance.

Any practical tips?

Businesses relying on AI will need to develop an AI system compliance strategy, which includes risk-assessing all AI systems, ensuring there is adequate documentation of testing, and implementing document retention policies to avoid any accidental destruction of potential evidence.

Winter 2022