EP Resolution on AI in Criminal Law and Policing
On 6 October 2021, the European Parliament (EP) adopted a resolution regarding artificial intelligence (AI) in criminal law and its use by the police and judicial authorities in criminal matters. The resolution was adopted by 377 to 248 votes with 62 abstentions.
According to the majority of MEPs, AI applications offer great possibilities for the field of law enforcement. The use of AI not only helps improve working methods in law enforcement and judicial authorities but is also useful in combating certain types of crime more efficiently (e.g., money laundering and terrorist financing, online sexual abuse, etc.).
MEPs are also aware of the potential for bias and discrimination arising from the use of AI applications. They noted that biases can be inherent in underlying datasets and that these biases tend to gradually increase, perpetuate, and amplify existing discrimination.
Another point of contention concerned the fact that many AI identification systems currently misidentify and misclassify racialised people, individuals belonging to certain ethnic communities, LGBTI people, children, the elderly, and women. In order to counter these problems, strong efforts should be made to avoid automated discrimination and bias. For this purpose, interdisciplinary research and input is required, including input from the fields of science and technology, critical race studies, disability studies, and findings from other disciplines.
Specific human oversight is necessary before operating certain critical applications, in order to avoid data leaks, data security breaches, and unauthorised access to personal data and other information related to critical applications. Law enforcement and judicial authorities should only use AI applications that adhere to the principle of privacy and data protection by design.
The resolution reinforces the fact that individuals working in the area of law enforcement and justice should not rely blindly on the seemingly objective and scientific nature of AI and always keep in mind that the results delivered by AI might be incorrect, incomplete, irrelevant, or discriminatory. It follows that decisions giving legal or similar effect always need to be taken by humans. Judicial and law enforcement authorities need to uphold extremely high legal standards and ensure human intervention. In this context, MEPs have called for a ban on the use of AI to propose judicial decisions. They cautioned that predictive policing cannot constitute the sole basis for an intervention, because it cannot answer the question of causality and cannot make reliable predictions on individual behaviour.
The resolution also calls for permanent prohibition of the use of automated analysis and recognition of human features in publicly accessible spaces. Critics also voiced concern over the use of private facial recognition databases, such as Clearview AI, by law enforcement actors and intelligence services.
Ultimately, MEPs are in support of a ban on AI mass-scale scoring of individuals, especially if used by law enforcement authorities and representatives of justice. In their opinion, this practice represents a loss of autonomy and a danger to the principle of non-discrimination and to fundamental rights.
The resolution from the EP comes after the proposed regulation from the European Commission laying down Harmonized Rules on Artificial Intelligence from April 2021 (→ see also eucrim 2/2021, p. 77) and after the EDPB/EDPS Joint Opinion on the AI Proposal that pointed the danger of the use of real time remote biometric identification systems by law enforcement authorities out (→ see also eucrim 2/2021, p. 77-78).