Should AI Play a Role in Modern Police Investigative Interviews?

Investigative interviewing represents the backbone of the criminal justice system, as the quality of these interactions directly dictates the accuracy and reliability of evidence gathered from witnesses, victims, and suspects. As AI in policing becomes a focal point for modernization, authorities are exploring how advanced technology might enhance or transform these critical inquiries.

The Evolving Role of AI in Policing and Investigations

For much of the 20th century, law enforcement frequently relied on accusatory methods that increased the likelihood of false confessions and distorted witness accounts. Human memory, rather than being a perfect recording, is a complex, reconstructive process vulnerable to external influences such as leading questions or media exposure. As the UK government launches a dedicated national AI centre for policing in June 2026, researchers are examining how machine learning can assist without compromising the integrity of investigative evidence.

In theory, AI in policing tools offer significant potential. Conversational AI agents could manage large-scale initial information gathering, while real-time prompts might guide officers to frame non-leading questions during later stages of an investigation. Furthermore, AI-enabled avatars are currently being tested to assist in training officers, providing a standardized approach to interviewing techniques. By automating routine tasks like transcription or data analysis, technology could theoretically allow investigators to focus on more complex decision-making.

However, significant risks remain regarding the impact of these tools on human cognition. Studies indicate that interacting with AI in policing systems can introduce misinformation, potentially leading to inaccurate recall. There is also a documented concern regarding automation bias, where investigators may defer to algorithmic outputs rather than relying on their own professional judgment. When systems appear highly confident, experts are more likely to trust the AI over their own assessments, which could propagate errors across entire investigative workflows.

The potential for contamination is particularly concerning when dealing with vulnerable witnesses, such as children or trauma survivors. Because the effects of AI-mediated interviews on jurors and the long-term reliability of evidence are not yet fully understood, researchers are calling for a systematic taxonomy of potential errors. As the legal community continues to emphasize the importance of established guidance like Achieving Best Evidence, there is a clear demand for innovation that aligns with rigorous scientific standards. Ultimately, while technology promises efficiency, the priority must remain the maintenance of evidence that stands up to the scrutiny of the courtroom.