Police officer is investigated for using AI to “create evidence” in the UK

UA British police officer is the target of a criminal investigation after suspicions emerged that he had used Artificial Intelligence (AI) to fabricate evidence in several cases.

“A criminal investigation has been opened due to an allegation of perversion of justice following the alleged use of an AI system by an officer to create evidentiary material in numerous cases,” said Derbyshire police, cited by The Guardian. “We are working closely with the Public Prosecutor’s Office with regards to any cases that may have been affected,” the official added.

For now, the investigation into the agent, who has not been identified, is still “in the early stages” and the types of cases that may have been targets of external manipulation nor the type of evidence that may have been created are not known.

“The officer involved has been removed from front-line duties pending the outcome of the investigation. No arrests have been made,” police added.

Meanwhile, the British Public Prosecutor’s Office, to the same media outlet, confirmed that an investigation is underway due to “the alleged use of AI by an agent”, but refuses any other type of comment, given the still initial state of the investigations.

The case arises after the London Metropolitan Police (known as the Met) installed software to monitor officers based on data to which the police have immediate access. Over the course of a week in April, the software revealed cases of minor infractions, such as teleworking rules, to cases of corruption and even rape. After that week, three officers were arrested for crimes of abuse of authority for sexual purposes, fraud, sexual assault, misconduct in the exercise of public functions and improper use of police systems.

More recently, however, the director of the National Police Chiefs’ Council’s Center for Police AI, Alex Murray, has instructed several police forces to stop using AI systems to prepare court statements and other tasks, saying the tools may not yet be sufficiently reliable.