AI is growing in usage, and it won’t be long before police and prosecutors start ramping up AI use in criminal cases. There are some novel uses where it might already be seeping into cases, but questions about how it can legally be used against you are not yet clear.
Law enforcement already uses AI in a few areas, such as “enhancing’ security camera photos of suspects and in license plate readers like Flock. These uses are somewhat limited, and things like this might not actually be admitted as evidence of a crime, but rather used to tip the scales toward police having “probable cause” to make an arrest. That being said, future attempts to introduce information produced by AI may not live up the scientific scrutiny needed for evidence admissibility, if it comes to it.
For help with your criminal case, call Overson & Bugden’s Salt Lake City, UT criminal defense lawyers at (801) 758-2287.
Using AI Evidence in Court vs. Using it for Probable Cause
One important distinction about how evidence is used is whether it is introduced in court to be used as evidence of guilt/innocence or whether it is used to establish probable cause to back up an arrest or charge.
- Actual evidence admitted in court must be obtained with a warrant or other legal means, and it must conform to evidentiary standards for scientific evidence, if applicable.
- Evidence used to establish probable cause does not necessarily need to be admissible, and it does not necessarily need to be scientific.
Examples
For example, DUI cases often use breath tests.
- Any breath test introduced as evidence must be part of a search incident to arrest (a warrant is not required, but there must be probable cause for the arrest). The machines used need to be properly calibrated, and the test needs to be given properly for the test results to be admissible.
- Presumptive breath tests (PBTs) used during a traffic stop are voluntary, do not need probable cause or a warrant, and do not need to be rigorously calibrated.
Because PBTs are not held to high standards, the evidence cannot be admitted as evidence of drunk driving in court; it is limited to being evidence of probable cause.
AI
When it comes to AI, there may be plenty of ways police can legally use AI to help them in an investigation. However, the “evidence” or information obtained through AI use might not be admissible evidence in court.
That being said, they may be able to still use it to establish probable cause.
Can AI License Plate Readers (LPRs) and Flock Cameras Be Used in Criminal Cases?
License plate readers (LPRs) are cameras that take pictures of license plates and run them through software to read the plate and compare it to databases for various investigative purposes. Companies like Flock have been ramping up sales and use of their LPRs, which can be installed on police cars or stationary locations.
How Flock Cameras and LPRs Use AI
The purpose of these LPRs is to track vehicles and reference where they were – all of which is typically done without warrants. These technologies do often rely on AI models to help them track locations and predict where vehicles may be going.
How These Are Used
Generally, these tools are used to locate vehicles and people, not as proof of any acts. Because of this, they do not always need to be introduced as evidence in court.
Any actual evidence introduced would probably be the photos of a car with a given license plate in a given location, which are certainly admissible.
Is that Legal?
There are arguments that taking pictures and running them through databases might constitute a search, meaning it could not be done without a warrant.
As Flock expands its technology, they seem to be aiming to do more intrusive things like scan for nearby devices’ Wi-Fi signals, which may be a search that needs a warrant.
Most of this has not been challenged or tested in court cases yet, so it is hard to say.
Can Police Use AI Images and Image Enhancement of Suspects?
For decades, things like police sketches have been used to help find suspects matching certain descriptions. When police have an actual photo of the suspect, this is usually better, but there is no legal issue with using a sketch. AI might serve a similar role.
How These Are Used
Police might try to use AI tools to create images matching a suspect’s description, then use them to find a suspect.
Famously, the Charlie Kirk assassination investigation saw police and other law enforcement in Utah using poor-quality images from security cameras “enhanced” by AI to try to locate the suspect in that case.
Is That Legal?
In these cases, the main purpose of using a sketch or AI-generated image would be to describe the suspect and try to find matches in public. They only need reasonable suspicion to stop someone and question them based on matching a picture of some kind – whether it be a sketch or an AI-generated or AI-“enhanced” image.
However, there may be ways that criminal defense lawyers can challenge this if it is used as an official identification in a photo lineup or something to that effect. Those procedures might not match the accepted standard required for lineups.
Can Police Use AI in Research?
Police and prosecutors more commonly use AI to help them with research. Whether they are looking for cases, laws, or scientific facts, they may be able to rely on these AI-driven results.
The problem arises when “hallucinations” occur and the police or prosecutors get wrong or made-up information. If they arrest you based on incorrect facts, we can challenge that in court just as we can with any other false information.
Can Prosecutors Use AI to Write Court Documents
Bar associations and courts across the country are wrestling with questions of how and whether AI can be used in legal research and preparing court documents. These issues must be compared against legal and ethical rules for lawyers, which cover prosecutors, too:
- It is unethical for lawyers to put forth cases that do not exist or do not say what they claim they say before a court. This could lead to sanctions, potentially including dismissed cases, fines, and disbarment.
- It violates the “duty of candor” for a lawyer to go before a court with half-true information and fail to correct false information or mistakes. This could come up in cases of AI-hallucinated citations as well as false information.
- If prosecutors have information that cuts against their claims – such as knowledge that the claims were researched from AI-driven sources – they may have an ethical obligation to disclose that.
Ultimately, the specific judge on the case has a lot of authority over these issues and would handle them on a case-by-case basis.
Call Our Criminal Defense Lawyers Today
For help with your case, call the Park City, UT criminal defense lawyers at Overson & Bugden at (801) 758-2287.
