Criminal Liability, Robotics and AI Systems (February 2021)

With Robotic and Artificial Intelligence ('RAI') systems being increasingly deployed across society, risks arise that those systems cause serious physical, psychological or economic harms. Against that backdrop, the LRC has considered whether, and if so how, criminal laws and penalties might be applied in such cases.

In its report, the LRC:

• Analyses both harms committed intentionally using RAI systems and those where serious harm results non-intentionally from RAI systems' operation, and highlights how particular challenges may arise in applying existing criminal law frameworks (in particular criminal negligence concepts) to the latter;
• Considers whether it is appropriate to apply criminal liability at all in relation to such non-intentional harms, or whether regulatory or civil powers and non-criminal sanctions may be a more appropriate way to promote safety without chilling innovation;
• Notes that alternative or supplementary approaches to attributing criminal liability for harms resulting from the operation of RAI systems may be necessary, and that a 'one size fits all' approach does not appear practicable; and
• Evaluates possible such alternative approaches, including those based on (a) legal personality for RAI systems, (b) certain new offences considered by the Penal Code Review Committee, and (c) the approach taken in legislation such as that governing workplace safety. 

A one minute quick guide to the alternative approaches considered in the report is available here.  

It is hoped that the report will assist policy makers in assessing where future legal and regulatory challenges may arise in seeking to hold someone criminally accountable for serious harms caused by RAI systems, and possible approaches to addressing those difficulties.

The full report can be found here.

The Attribution of Civil Liability for Accidents Involving Autonomous Cars (September 2020)

The much-anticipated mainstream adoption of autonomous cars is fast becoming reality. While such “self-driving” technologies are predicted to reduce drastically the number of road-traffic accidents, incidents will inevitably still occur. Questions therefore arise as to who should be liable for those accidents where they result in injury or harm, and on what legal basis.

In that context, this report:

• Identifies how complex autonomous driving technologies can complicate the process of: identifying what caused an accident and determining who should be liable for it; establishing that party’s liability; and assessing any relevant defences;
• Surveys the regulatory steps taken by key overseas jurisdictions (including the EU, US and Japan) to adjust their laws to accommodate the adoption of autonomous vehicles on public roads; and
• Highlights the challenges that arise in seeking to apply each of the main existing liability frameworks - negligence, product liability and 'no-fault' liability - in the context of autonomous cars (click here for a one-minute quick guide to the key challenges under each framework). 

It is hoped that the report will assist policy makers in assessing the best future legal and regulatory scheme for Singapore, so as to encourage innovation and promote the mainstream adoption of autonomous vehicles, while also ensuring those harmed in accidents obtain effective redress.

The full report can be found here.

Applying Ethical Principles for Artificial Intelligence in Regulatory Reform (July 2020)

There is a growing consensus that the increasing deployment of AI systems across society, as well as bringing benefits, must also be human-centred and built on strong ethical foundations.

That, in turn, has implications for the laws and regulatory interventions that enable, guide and constrain how and where AI is deployed.

This report identifies issues that law and policy makers may face in promoting ethical principles when reforming laws and regulations to adapt to AI, and provides examples of human-centred approaches that could be taken to address these.

Specifically, the report discusses the following ethical principles:

The report does not seek to advance specific means or level of intervention, which will necessarily vary depending on the technology and sector in question. Its purpose, however, is to provide a framework for broader consideration and discussion on the best means to achieve human-centred, ethical norm-making and calibration of regulatory responses regarding AI.

The full report can be found here.

Rethinking Database Rights and Data Ownership in an AI World (July 2020)

Data and databases are central to the ongoing development and use of AI-enabled systems and technologies. At the same time, concerns remain regarding individuals' control over the use of their data by third parties.

Against that backdrop, this report considers whether key data-related laws in Singapore currently operate effectively to promote the beneficial production of, and access to, databases, while also protecting individual rights.

In particular, the report addresses:

• who controls or has rights over the ‘big data’ databases that underpin AI technologies, including whether new, standalone intellectual property protections for databases are needed; and
• how to ensure that those who contribute data to databases retain appropriate control over and/or access to that data, including whether to create a new right of ownership of data.

Summary of Recommendations

• As regards databases, that:
  • Amendments are made to copyright law to expressly protect ‘computer-generated works’ that have no human author.
  • Clarification is given through guidance or subordinate legislation regarding: a) how existing ‘compilation rights’ apply to electronic databases; and b) how effective records of authorship of databases can be maintained where several people have contributed to the database’s creation.
  • Creation of a new standalone intellectual property right for databases is not required nor necessarily beneficial in the Singapore context.
• As regards control of data, that:
  • For personal data, existing and incoming data protection laws (including in particular data portability) currently provide individuals with sufficient control over their data. 
  • For non-personal data, consideration be given to introducing a right akin to data portability. 
  • A right to ‘own’ data as property is not required, nor would it necessarily be practicable or beneficial.

The full report can be found here.