AI Regulation in Forensics: Challenges and Obligations

Forensics under the Artificial Intelligence Act

The present decade has seen the widespread deployment of AI systems across nearly all areas of human activity. The prospect of mass commercialization of AI systems has greatly contributed to an increase in funding for AI research. However, alongside these opportunities, there are significant risks associated with AI technology. Many AI systems are complex, making their decision-making processes difficult to explain, often referred to as black-boxes.

Artificial Intelligence in Forensics

One of the first examples of forensic AI systems was an artificial neural network developed between 1993 and 1995 for scoring polygraphic signals. The development of forensic AI applications has evolved alongside the general progress in AI research. Recent literature suggests a growing interest in the use of AI in forensics, such as:

Determination of sex or age of individuals based on biological samples like teeth, saliva, bones, or shoe prints.

Artificial Intelligence Act

The general goal of the Artificial Intelligence Act is to lay down a uniform legal framework for the use of AI systems in the Union while ensuring a high level of protection for health, safety, and fundamental rights against the harmful effects of AI systems, thus supporting innovation.

Forensics as a High-Risk Area

Annex III of the Act identifies two areas of high-risk applications related to forensics: law enforcement and administration of justice. The scope of the law enforcement area includes AI systems intended for use by law enforcement authorities or on their behalf. Specific use cases for law enforcement include:

AI systems designed to assist in criminal investigations.

Obligations of Forensic Experts as Deployers

General requirements for high-risk AI systems include:

Implementation and maintenance of a risk management system throughout the entire lifecycle of the system.
Preparation of technical documentation demonstrating compliance with the Act’s requirements.
Provision of necessary information to relevant authorities for compliance assessment.

Penalties for Non-Compliance

Non-compliance with the Act may result in administrative fines as specified in the Act, along with other penalties and enforcement measures that Member States may impose. The choice of authorities to enforce fines will also be determined by the Member States.

Conclusions

The introduction of the AI Act represents the most significant step toward AI regulation to date. It is a comprehensive response to the risks associated with AI, particularly in forensics. The Act imposes obligations on operators of AI systems based on the risks inherent to these systems, and it is suggested that forensics be classified as a high-risk area given the growing impact of AI technology in this field.