Ai In Policing: Effective Or Dangerous?

Artificial intelligence (AI) is increasingly being used by law enforcement agencies to improve efficiency and accuracy. AI can automate back-office tasks, analyse data, and assist with suspect identification. It can also be used for crowd control and surveillance, facial recognition, and scanning video footage for anomalies. While AI can be a valuable tool for law enforcement, it is essential to consider the ethical implications and potential risks. The responsible use of AI in law enforcement aims to balance personal privacy and public safety, with ongoing conversations between stakeholders necessary to navigate the complex task of institutionalising it.

AI and personal privacy vs public safety

AI technologies are currently being used to assist law enforcement agencies in a variety of ways, from automating back-office tasks to accelerating investigations and processing evidence. While AI can undoubtedly help law enforcement perform their duties more efficiently and accurately, its implementation raises important questions about personal privacy and public safety.

AI systems often rely on large amounts of personal data for training and inference, which can lead to privacy concerns and the potential for misuse or unauthorized access to sensitive information. For example, the use of AI-driven facial recognition technology has faced scrutiny due to instances of misidentification and inaccuracies, resulting in wrongful arrests and infringements on individuals' rights. The combination of CCTV cameras in public spaces with facial recognition software can also create a tool that is highly invasive of privacy. Additionally, AI tools trained with data scraped from the internet may memorize personal information about individuals, as well as relational data about their family and friends, which could be used for anti-social purposes.

To address these concerns, it is crucial to develop AI models and algorithms that inherently respect and protect individual privacy. This can be achieved through techniques like homomorphic encryption, secure enclaves, and privacy-preserving machine learning. Regulatory frameworks, such as the General Data Protection Regulation (GDPR) or the California Consumer Privacy Act (CCPA), provide guidelines for organizations developing and deploying AI systems to protect personal information. However, the onus is often on the individual to understand and exercise their data rights, which can be challenging and time-consuming.

As AI continues to evolve and become more integrated into law enforcement, ongoing conversations between stakeholders, including community representatives and criminal justice agencies, are necessary to navigate the trade-offs between personal privacy and public safety. This includes discussing the ethical implications of expanded surveillance and investigative capabilities and ensuring that the benefits of AI in law enforcement are not outweighed by the potential risks to civil liberties.

In conclusion, while AI has the potential to revolutionize law enforcement and improve public safety, it is essential to prioritize the protection of personal privacy through strategic and thoughtful regulation, ethical considerations, and the development of privacy-preserving AI techniques.

AI for faster, safer, more accurate suspect identification

The use of AI in law enforcement is no longer a tool of the future but a present-day necessity. As technology continues to advance, law enforcement agencies must also be equipped with the latest tools to help them perform their duties more efficiently and accurately. AI can be invaluable in increasing productivity, reserving man-hours, optimising workflows, and allocating more resources towards human-centric tasks that cannot be automated.

One of the most notable applications of AI in law enforcement is in suspect identification. Traditional methods of identifying suspects rely on eyewitness accounts and forensic evidence, which can be time-consuming and challenging. With the advent of AI technology, suspect identification has been revolutionised, enhancing accuracy and efficiency. AI can swiftly analyse vast amounts of data, including video, audio, image files, log files, and social media posts, which often make up a large portion of the evidence.

AI-enhanced scanning and facial recognition technology can be used to compare perpetrators in video and photographic evidence against booking databases of known offenders. This can be further refined by filtering potential matches with key identifiers such as gender, age, height, hair colour, eye colour, and ethnicity. For example, Veritone IDentify has helped a Southern California Police Department with a database of 20,000 images to identify suspects faster and more accurately, leading to more case resolutions and saving valuable resources, time, and money.

AI can also assist in redacting specific items, such as heads or license plates, from audio, video, or image-based evidence, which is often a time-consuming and expensive process for agencies. This not only improves investigative workflow but also helps maintain the privacy and safety of individuals. While AI offers transformative benefits, it should be used to complement rather than replace traditional methods, ensuring a balanced approach that improves public safety and increases the efficiency of crime prevention efforts.

AI for redacting sensitive information

Artificial intelligence (AI) is a present-day need for law enforcement agencies, helping them perform their duties more efficiently and accurately. AI can be used to automate back-office tasks, such as the manual redaction of sensitive information, which is time-consuming and expensive.

AI-infused tools like Veritone Redact can detect and redact specific items (e.g. faces, license plates, credit card numbers) in a quick and cost-effective manner. After redacting their audio, video, or image-based evidence, agencies can download the information to their local computers for further processing. AI redaction software is more accurate than manual redaction, as it reduces the risk of human error and ensures that all sensitive information is properly redacted.

AI redaction tools can handle various document types and large datasets with a higher degree of accuracy, ensuring that any confidential information is blurred to unauthorized users. These tools help companies stay compliant with evolving data privacy laws, including HIPAA, GDPR, CCPA, and more. They can easily adapt to increasing data loads and other sensitive data types, making them ideal for growing businesses across different fields.

AI-based redaction uses machine learning algorithms to automatically detect and remove every instance of sensitive and personally identifiable information (PII) from documents, images, and other digital content. It combines optical character recognition (OCR), natural language processing (NLP), and contextual pattern analysis to detect and remove data with high precision. This helps organizations maximize information security, ensure compliance, and increase operational efficiency.

AI for processing unstructured data

AI solutions are a present-day necessity for law enforcement agencies (LEAs) to improve efficiency and accuracy. AI can automate back-office tasks, reserve man-hours, optimise workflows, and allocate more resources towards human-centric tasks. One such example is Veritone's AI solutions, which automate the back-office tasks that take dozens of employees and hundreds of hours to complete.

Law enforcement agencies often have to deal with unstructured data, which is time-consuming to manually search through and analyse. This includes video, audio, text messages, witness statements, and social media information. As a result, large amounts of digital evidence may go unreviewed. AI can be used to process this unstructured data more efficiently. For example, Veritone Illuminate assists with analysing electronically stored information (ESI), including unstructured data. Similarly, Cognyte's NEXYTE decision intelligence platform enables the automation of data collection and analysis, reducing the time analysts spend on tedious tasks. It can pull unstructured data from disparate sources, including natural language processing (NLP), OCR data extraction, speech recognition, and video.

AI-infused tools can also be used to detect and redact specific items in a quick and cost-effective manner. For instance, Veritone Redact can be used to redact audio, video, or image-based evidence. Furthermore, AI can be used to identify trends and support law enforcement strategies, as well as provide instant, actionable intelligence to officers. CrimeTracer, an advanced analytics tool, generates immediate investigative leads for field officers and detectives. By using this tool, the Oakland Police Department saw a significant decrease in violent crimes, including a 50% decrease in shootings, a 42% drop in homicides, and a 38% decline in robberies over five years.

While AI can greatly assist law enforcement, it is important to consider the trade-offs between personal privacy and public safety as AI applications evolve to provide expanded surveillance and investigative capabilities.

AI for crowd control and surveillance

AI is increasingly being used to assist law enforcement agencies in their duties. AI technologies can help law enforcement agencies to work more efficiently and accurately, reserving man-hours, optimising workflows, and allocating more resources towards human-centric tasks.

AI can also be used to detect potential hazards in crowded environments. For instance, firearm detection technology can recognise the presence of a weapon within seconds, allowing authorities to intervene before an incident occurs. AI surveillance systems can also detect fires, accidents, or security breaches, triggering alerts and expediting response times. Thermal imaging technology can detect heat, revealing people, vehicles, or objects, even in low-visibility conditions. Heat mapping can highlight unusual temperature spikes, and predictive modelling can be used to preempt crises.

AI-powered tools can also be used to manage crowds at events. For example, crowd segmentation involves dividing an audience into smaller, manageable groups based on location, behaviour, or ticket type, which helps to maintain control over large crowds. Wearable technology can also be used to enhance crowd control and safety, with security personnel equipped with devices that provide real-time alerts about crowd density, potential hazards, or security breaches.

Frequently asked questions