AI in Law Enforcement: A Risky Experiment?

Law enforcement agencies are grappling with how to use AI tools safely, ensuring accuracy, privacy, and professionalism. However, this is no easy task. A recent case highlights the potential pitfalls.

The Case of the Immigration Officer

An immigration officer used ChatGPT to assist in writing a report. The officer provided only one sentence and a few pictures, a practice experts warn against. This approach is considered risky, especially in high-stakes situations.

The Department of Homeland Security has not yet clarified whether they have rules regarding AI usage. Additionally, they have not released the body camera footage referenced in the report, leaving the public in the dark about the incident.

Lack of Regulations

Experts note that few police departments have established rules for AI use. Some departments prohibit AI from assisting in reports involving use of force, as courts require precise details about the incident and the officer's mindset—something AI may not reliably provide.

Privacy Concerns

Using AI also raises privacy issues. If the officer used a public version of ChatGPT, the images may have entered the public domain, potentially accessible to malicious actors.

Playing Catch-Up

Experts argue that police departments often lag behind technological advancements, waiting for problems to arise before implementing rules. A proactive approach—understanding risks first—would be more effective.

State-Level Initiatives

Some states, such as Utah and California, are taking the lead by requiring AI-assisted reports to be labeled. This transparency could be a step in the right direction, helping the public identify when AI has been involved.