FB pixel

Can criminal investigations rely on AI?

Can criminal investigations rely on AI?

By Ján Lunter, CEO of Innovatrics

Incorporating AI technologies in criminal investigation can expedite the process, generate high-quality leads, and enhance law enforcement efforts. However, as with any technological advancement, it raises important questions about ethics and the potential for misuse.

The most crucial aspect of criminal investigation is time. The faster the police can obtain leads and information, the more efficiently they can solve crimes, prevent escapes, and protect potential victims. Homicide detectives and experts always tell us that if you can’t find a lead within the first 48 hours, the chance of solving the case decreases exponentially.

This is where AI comes into play. Let’s explore the possibility that criminal investigation will increasingly rely on AI in the future and consider whether this development could be a double-edged sword.

The AI applications for investigations

Firstly, AI-driven fingerprint analysis can quickly identify latent prints, saving hours or days compared to manual forensic examination. Obviously, the accuracy of these identifications still requires validation by trained forensic examiners, ensuring that AI remains a tool rather than a decision-maker.

Similarly, facial recognition technology can help identify individuals from photos or security camera footage, reducing the time needed for apprehension or further interrogation. AI advancements like forensic cameras for non-destructive fingerprint collection can also expedite on-site investigations.

Video analysis can help law enforcement identify suspect potential person meetings near crime scenes and collect valuable evidence for use in court. There is even a team of researchers in Malaysia actively working on the development of AI software designed for CCTV cameras to diminish street crime incidents.

The researchers believe the software boasts a range of capabilities, including the ability to discern whether an individual in the video is carrying a weapon, scrutinize for signs of “aggressive actions” exhibited by suspects, and promptly alert law enforcement authorities in the event of suspected criminal activity. Furthermore, AI contact analysis and graph databases assist in recognizing obscure connections between suspects, crime scenes, and victims.

Despite all this promise, it is important to stress once more that AI technology does not replace human decision-making and expertise. Instances of victims being falsely accused serve as reminders that faults lie in the investigative process, not the technology itself. AI acts as an invaluable assistant, but human specialists must make the final determinations.

The risks of false arrest

AI, particularly facial recognition technology, presents certain risks concerning identification. In some cases, false arrests have occurred due to misidentifications made by AI-driven systems.

Recently, Randal Quran Reid was wrongfully arrested due to misuse of AI. The recognition technology used in the case developed by Clearview AI, Inc., has since faced criticism despite over one million searches conducted. As the saying goes, “one wrongful arrest is one too many,” and there is public concern over the potential for innocent individuals to be detained based on technology use errors.

This just highlights how AI should serve as a starting point for investigations, requiring further verification based on additional evidence and factors. The initial result should be the beginning of the investigation, with law enforcement using various factors to determine the correct person’s identity rather than jumping to conclusions.

Transparency and accountability

Current policies are slowly being developed across the U.S. to address AI-related concerns, particularly facial recognition technology. One notable development is the inclusion of a disclosure stipulation explicitly related to facial recognition technology leads. This ensures that any leads generated during an investigation are discoverable by defense attorneys, which is a crucial step toward transparency and accountability.

It is a fundamental principle in criminal investigations that all actions must be discoverable. This includes AI-generated leads, which does raise the question of whether a specific stipulation is even needed for facial recognition technology when it should be treated like any other investigative lead. However, the absence of this kind of caveat implies that some police departments may not routinely disclose the use of facial recognition technology, which is morally questionable.

Improving the technology

AI can also help improve the technology being used, leading to less bias and more reliable results. The AI in its current form relies heavily on large datasets to “learn” how to distinguish persons, objects etc. For some uses there is not enough data, e.g. in latent fingerprints (the fingerprints left on the crime scene). For these, we are actually able to use AI to create synthetic latent fingerprints in order to help algorithms discern these partial clues better.

Similar approach can be used to amend datasets with underrepresented minorities, age groups and so on to prevent the most obvious biases of facial recognition technology.

Final thoughts

Using AI in criminal investigation holds significant promise for speeding up the process, generating high-quality leads, and making law enforcement more efficient. However, the technology is clearly not without its challenges, including the potential for false arrests and concerns about transparency and accountability.

AI should always be regarded as a tool in the hands of human experts, and its limitations and potential errors must be acknowledged. The future of criminal investigation may rely more on AI, but whether this is a good or bad thing depends on how well society balances technological advancement and ethical considerations.

About the author

Jan Lunter is Co-founder and CEO of Innovatrics, which has been developing and providing fingerprint recognition solutions since 2004. Jan is an author of the algorithm for fingerprint analysis and recognition, which regularly ranks among the top in prestigious comparison tests (NIST PFT II, NIST Minex). In recent years he is also dealing with image processing and the use of neural networks for face recognition.

DISCLAIMER: Biometric Update’s Industry Insights are submitted content. The views expressed in this post are that of the author, and don’t necessarily reflect the views of Biometric Update.

Article Topics

 |   |   |   |   |   | 

Latest Biometrics News


What to do if certificates for passive authentication fail

By Ihar Kliashchou, Chief Technology Officer at Regula Electronic documents are praised for their top-notch security, mainly due to RFID chip…


Biometric passports get refresh in Indonesia, face hurdles in Lebanon and Kenya

Nations across the global south are looking to biometric passports as the next generation of travel documents. Indonesia will mark…


Biometrics, electronic devices and identity credentials converging

Biometrics in electronic devices and ID documents to support digital identity are a major theme of the week’s top stories…


SITA wraps up acquisition of Materna IPS

SITA reports it has completed all necessary regulatory and legal procedures and finalized its acquisition of Materna IPS, a provider…


Payface lands new retail biometric payments deal in Brazil

Brazilian face biometrics payments startup Payface has clinched a deal with supermarket chain Ítalo. Ítalo Supermercados, based in the southern…


EU to fund digital programs with €108m, including digital identity

The European Union has issued a new call for funding within the Digital Europe Programme (DIGITAL), allocating over 108 million…


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Most Read This Week

Featured Company

Biometrics Insight, Opinion

Digital ID In-Depth

Biometrics White Papers

Biometrics Events