Imagine a robbery takes place in a shop late at night. There are no people around to see what happened. The only thing that captures the incident is an AI powered security camera. It records the suspect, recognises unusual behaviour and automatically alerts the police. Later, that recording becomes one of the main pieces of evidence in court.
This brings up an interesting question. Can AI actually be considered a witness?
At first, the answer seems obvious. A witness is usually a person who saw or heard something and can explain what happened in court. They can answer questions, take an oath and be cross examined by lawyers.
Artificial intelligence cannot do any of those things. It cannot remember events the way humans do and it has no understanding of truth or lies. It simply collects information and processes it according to the way it has been designed.
Even though AI cannot testify in court, the information it collects can still be very valuable.
Today, AI is already being used in many situations that involve legal investigations. Security cameras can identify suspicious activity. Self driving cars record everything that happens before an accident. Smart home devices store voice recordings and activity logs. Banks use AI to detect fraud, while facial recognition systems help identify suspects. These technologies are becoming part of everyday life, which means they are also becoming part of the legal system.
However, using AI as evidence is not without problems.
One major concern is accuracy. AI systems can make mistakes just like humans. A camera might identify the wrong person because of poor lighting, or an algorithm might produce incorrect results because it was trained using biased data. If the evidence is wrong, it could affect the outcome of an entire case.
Another issue is transparency. Some AI systems make decisions through complex algorithms that are difficult to understand, even for the people who created them. In court, evidence should be clear enough for judges and lawyers to examine and question. If nobody can explain how an AI reached its conclusion, trusting that evidence becomes much more difficult.
Responsibility is another challenge. If an AI system makes a mistake, who should be held accountable? Is it the company that developed the software, the organisation using it, or someone else? Current laws do not always provide clear answers.
Privacy also becomes an important issue. Many devices collect personal information every day without people thinking much about it. Smart speakers, fitness watches and home security
systems all store data that could later be used during investigations. While this information may help solve crimes, it also raises questions about how much access authorities should have to our private lives.
So, can AI be a witness?
Legally, the answer is no. A witness must be able to give testimony and answer questions before a court. AI cannot do that because it has no personal knowledge or legal responsibility.
What AI can do is provide digital evidence. Courts may use recordings, data and analysis produced by AI systems, but that evidence still has to be checked carefully before it can be relied upon.
As AI continues to develop, the legal system will also need to change. New rules will be needed to ensure that AI generated evidence is accurate, reliable and used fairly. Technology is advancing quickly and the law has the difficult task of keeping up.
AI is unlikely to replace human witnesses anytime soon. Instead, it is becoming another source of evidence that can help investigators and courts understand what happened. Whether that evidence should always be trusted is a question that lawyers, judges and governments will continue to debate for many years.
Written By: –
Hiruni Keshara De Joedth
(University of Kelaniya)
Designed by: –
Rtr. Chamodya Anushani
(Senior Blog Team Member 2026-27)
