By: Paige Cerulli
How artificial intelligence is changing criminal justice
Artificial intelligence (AI) has rapidly transformed many industries, from healthcare to education and beyond. AI is also changing the criminal justice field. Many criminal justice professionals are leveraging this technology and benefitting from its numerous perks, but AI has some negative effects that are impacting the industry, too. As AI rapidly advances, it’s already shaping criminal justice and is poised to continue to impact the field for years to come.
In this Article
How criminal justice professionals are using AI today
Many criminal justice professionals already use AI to do their jobs more efficiently and accurately. Bridget Lowrie, a criminal defense attorney, teaches in the Sociology, Anthropology and Criminal Justice Department at Montgomery College. “AI is not new to criminal justice,” Lowrie said. “Using technologies like license plate readers for finding probable cause to stop vehicles, spelling and grammar check for police reports, and facial recognition have been around for many years before what we think of as the modern invention of artificial intelligence in the last few years.”
Lowrie added that tools like ShotSpotter—which uses sensors to capture sound waves to identify potential gunfire and its location—also use AI, though these tools have been in existence before recent AI innovations. “What we are seeing now is that these tools can use modern AI to potentially have even more accuracy and time savings in the future,” Lowrie said.
Using AI for good
The Center for Justice Innovation describes the strategic use of AI to identify disparities in how the legal system uses fines and fees. AI analyzed data on the issuance of fines and fees in Jefferson County, Alabama, finding that low-income and Black communities were disproportionately impacted by fines. Such information can be used to raise awareness, identify human bias and work to correct unfair treatments.
The Center for Justice Innovation also notes that AI can be used to analyze data, monitor the impact of existing policies and programs, and help communities work to strengthen those programs. AI can also help reduce administrative tasks, freeing up criminal justice professionals to provide direct support.
There are many, many other practical applications of AI in the criminal justice field, including:
- Automated fingerprint identification systems:
- AI can save professionals time by scanning fingerprint databases and identifying matches. It can help speed up match identification, which can be key for professionals working on time-sensitive cases.
- Facial recognition technology:
- Facial recognition technology gives criminal justice professionals another tool to help identify persons of interest. The technology saves time and may provide leads when fingerprints aren’t available.
- Emergency response tools:
- Emergency response services can use AI systems to prioritize emergency calls based on their urgency. These systems can analyze information and help with informed decision-making, such as determining which officers should respond and which routes they should take.
- Crime analysis tools:
- Crime analysis tools can analyze massive amounts of crime data to better understand crime patterns. With that information, police departments can better and more strategically allocate their resources to reduce crime, even when working with limited budgets.
- Forensic analysis tools:
- Forensic analysis tools are capable of processing extensive amounts of data to finding information and identify patterns that help professionals examine and understand crime scenes.
- Cybercrime tools:
- Cybercrime investigators are using AI tools to identify and investigate digital crimes. These tools help prevent and investigate crimes such as online fraud, hackers, and more.
- Courtroom tools:
- Some courts have begun using AI tools to support reporting and transcription needs. Judges and lawyers may also use AI to analyze and summarize legal documents.
“AI allows criminal justice systems to make more data-driven decisions across the board,” says Lowrie. “It improves efficiency in processing information and helps agencies stretch limited resources.”
Dr. Heidi S. Bonner, Department Chair and Professor for the Department of Criminal Justice and Criminology at East Carolina University, noted that AI’s greatest impact on criminal justice so far is its ability to condense massive amounts of unstructured data into more digestible summaries.
“Language Multimodal Models have changed criminal justice less by replacing people and more by reshaping how information moves through the system,” Bonner said. “They can instantly summarize jail calls, body-cam footage, investigative files, and court documents, which speeds up policing, prosecution, defense work, and even judicial decision-making.”
As a result, professionals often start their work with an AI-generated narrative, rather than working with raw evidence. That shift could influence how cases are framed and how risk or intent is interpreted, Bonner explained. “Language Multimodal Models make it possible to process far more data far faster, but they also introduce vulnerabilities around accuracy, authorship, privacy and due-process protections that the system is still scrambling to govern.”
New challenges AI brings for criminal justice professionals
AI offers many benefits to the criminal justice industry, but it also has drawbacks and poses new challenges.
AI and human biases
AI can—and has—reflected human biases, which could skew its use in criminal justice applications. Lowrie referenced a study on AI chatbot biases in medical diagnoses. The study, published in the National Library of Medicine in 2023, revealed that AI chatbots provided different recommendations based on a patient’s gender, race and ethnicity.
“Racial, gender and socioeconomic stereotypes already exist in our society, and AI often reflects them,” Lowrie said. “As law enforcement agencies struggle to recruit new officers and budgets get cut, there will be pressure to rely more and more on these technologies as a cost-cutting measure, and the profession as a whole needs to be aware of these biases in AI to maintain public trust.”
Numerous factors can contribute to AI bias. AI models can reflect the bias of the developers that created them, and AI can reflect a bias caused by the data that the model is trained on. When AI is used for criminal justice, those biases could lead to incorrect or skewed information. If professionals are relying on that information to convict individuals or to issue tickets and fines, those actions could be taken incorrectly and unfairly, which could erode public trust in AI and the criminal justice system. While AI can absolutely help streamline processes and improve efficiency, it’s essential that developers take steps to mitigate bias, especially when implementing AI in criminal justice applications.
AI inaccuracies
Bonner explained that AI can produce a reasoned narrative, which humans can trust too much. “It’s a subtle concern, but Language Multimodal Models can persuade, not just answer,” she said. Individuals need to stay aware of this potential persuasion, especially when using AI-generated responses and information in the criminal justice field.
AI can also hallucinate. “Agencies need verification workflows, which many don’t yet have,” Bonner said. “Language Multimodal Models also require folks who understand data literacy, model limits and procedures for documenting AI involvement, and most agencies don’t have this training infrastructure yet.”
The CSI effect
Media, including popular television shows like CSI: Crime Scene Investigation (often shortened to CSI), have created unrealistic expectations and assumptions around the use of technology, including AI. Lowrie describes the “CSI effect,” which led to a perception that law enforcement professionals could remove a tiny grain of sand from a crime scene, process it in a computer and have that computer produce a fingerprint, DNA, driver’s license photo, home address, social security number and more details about the suspect.
“Law enforcement will struggle to explain to juries that AI has not created super cops that can know where crime is going to happen before it does and instantly download the contents of a suspect’s phone to know if they committed the crime,” Lowrie said. “Personal privacy, constitutional safeguards and human review are still necessary parts of the criminal justice system.”
Privacy concerns
When it comes to AI use in surveillance, questions around civil liberties arise. The criminal justice system must address how to strategically and appropriately use AI without infringing on privacy and civil rights. The use of AI is largely unregulated so far, so criminal justice departments must walk a fine line in how they use the technology.
Non-ethical use
AI has given individuals new capabilities when it comes to deepfakes, surveillance, hacking and more. Unethical use of the technology can create a negative public perception.
The truth is, AI can be used for good and for bad. As criminals use it in new ways, the criminal justice system is quickly trying to identify these methods and use AI to stop them.
Where the criminal justice field will be in a decade
Lowrie compared AI’s invention to the invention of the printing press, noting that people were initially horrified and began destroying the printing press because they were worried about its capability to rapidly spread misinformation, costing scribes their jobs. “In some ways they were right, but as a whole, our society is still better for the technology,” Lowrie said. “I think long term we will hopefully see the same thing with AI. There will be some short- and long-term losses, but hopefully, in ten years, our society is better for it.”
Bonner said that AI will almost certainly become even more embedded throughout the criminal justice system. While it will speed up investigations and case processing, AI also carries risk. “Whether AI ends up doing more harm or good depends almost entirely on whether the system builds transparency, oversight, and limits now rather than after the damage is done,” Bonner said. “With strong guardrails, AI could ease workloads and bring more consistency to decisions. Without them, it’ll simply speed up the inequities that already exist.”
AI may bring new challenges and issues to criminal justice, but Lowrie also believes it’s here to stay. “Despite real concerns, it is here, and now our concern needs to be about using it responsibly rather than pretending it doesn’t exist,” Lowrie said. “I’m incredibly lucky to be in higher education and working with students who are very excited about criminal justice. Most of them use AI in all kinds of ways in their lives and have great ideas and insights for how to use this technology.”