Harnessing AI for justice: Transforming war crime investigations

Yodunika Seneviratne
Feb 21
6 min read

Yodunika Seneviratne | 24 January 2025

The image shows a glossy statue of justice holding scales. The figure is bisected with glowing, gold, lines evoking digital networks. — AI image of Justice and scales, generated with Canva AI photo generator

“What AI and machine learning allow you to do is find the needle in the haystack.”

~Bob Work

War crimes, crimes against humanity, and genocide are internationally recognized as grave violations. International law stipulates that violence inflicted upon civilians must dealt with using available legal instruments and that the perpetrators should be brought to justice. The International Criminal Court (ICC) is a key institution for the pursuit of justice in these crimes. However, for ICC to open an investigation, the prosecutor must determine, after a preliminary examination, that the alleged crimes meet the threshold of "sufficient gravity." Once an investigation is initiated, the prosecutor's office usually dispatches investigators and other personnel to gather evidence which are in diverse forms and large amounts. This evidence will then be carefully analyzed to build a case, ensuring that the evidence meets the standards of admissibility in court.

Given the complexity of the cases and many administrative and bureaucratic hurdles, The ICC has been criticized by many actors, including both state actors and non-state actors like Amnesty International. Many of these criticisms focus on the delayed delivery of justice due to the complexity of the investigations, the challenges in accessing evidence and crime scenes and the high standards of proof. For instance, the war crime investigations in Uganda which started in 2004 were concluded only in 2023.

The sheer amount of data related to war crimes can make the processing time-consuming, complex and tedious. This is where Artificial Intelligence (AI) can play a pivotal role. In order to process, aggregate, disaggregate and analyze large amounts of data, war crime investigators can utilize infrastructure made with AI. For example, OTPLink an application for evidence submission for ICC, incorporates advanced AI technologies to streamline data processing.

Recent advancements in AI have transformed sectors ranging from education and health care to security and personal assistance. Exploring how AI can be harnessed to serve justice is a vital step forward. Having said that, recent developments in the warfront show the utilization of AI-integrated weapons bring tactical advantages that bring human costs. Critics warn that delegating the decision-making about life and death to AI can result in catastrophic outcomes as AI is making decisions solely on the facts and mostly historical data, it lacks the empathy that should be part of the decision-making.

In the protracted war between Russia and Ukraine, there have already been attempts to use AI-integrated drones and weapons which can identify targets, operate in swarms and adapt to new threats. With these, we are delegating the decision-making power to a robotic device which could be deadly and ruthless. However, we can use AI itself to mitigate these risks both on the battlefield and in post-war situations. As seen from its capability, AI could be a double-edged sword, as both a potential cause of harm and a solution. If applied ethically and responsibly it can be a beneficial apparatus to instigate justice for crimes that happened in conflict zones.

For example, satellite imagery has been around for some time making the skies above us a pair of eyes which can see everything. However, it is often difficult to find evidence using satellite imagery, as the material generated is so vast. Therefore, it is helpful that AI-integrated object detection models have the capacity to process satellite images to detect war crimes such as bombed hospitals, mass graves etc. Amnesty International has used this AI-integrated method to detect mass graves during the Rohingya crisis (2016-2017).

Moreover, when someone uses the internet it leaves traceable data and activity of that person, a so-called digital footprint. The digital footprint of war crimes can also be left behind in the form of social media posts or videos by witnesses or sometimes even perpetrators. AI tools such as PhotoDNA can be used to find, for instance, children who are victims of sexual abuse by tracing the location and stopping the redistribution of illegal images. These tools can also verify the authenticity of online content. AI-integrated software like TruePic can identify the authenticity of online content by analyzing its origins and any suspected tampering. Ensuring the authenticity of evidence is critical, as AI-generated or altered content could mislead the investigation.

Furthermore, war crime investigations often require connecting many different incidents to one another to unravel the patterns and motives behind them and find possible leads. In that regard, AI can play a crucial role in identifying critical patterns that can be missed by human investigators. AI can analyze data from different data sources and map incidents and violence. Moreover, advanced machine learning technologies can also help to provide a forecast about future conflicts, using historical and real-time data to identify patterns and establish connections. For example, an International, impartial non-profit organization called Armed Conflict Location and Event Data Project (ACLED) provides open-access tools for analyzing and visualizing conflict trends worldwide.

Testimonies by victims and witnesses are a cornerstone in war crime investigations, but analyzing thousands of testimonies and interviews is a daunting and resource-intensive task. This arduous process can be streamlined and made more efficient with AI-integrated tools, such as AI-integrated speech-to-text and sentiment analysis like Transcom, which has the ability to conduct speech recognition, preprocessing and sentiment analysis in speech. This can assist in identifying emotional stress, inconsistencies and patterns across interviews.

Apart from that, AI can be used not only in investigation and interrogation but also in case management. AI can streamline the legal investigative process by properly organizing large volumes of evidence and documents. AI tools like Relativity or CaseMine can help prosecutors review evidence faster by using algorithms to see patterns and build chronologies and by linking the events to specific individuals or parties. These new AI advancements can thus save a lot of time and resources, allowing the courts to process cases more efficiently.

The potential of AI in war crime investigations is increasingly being recognized. ICC Prosecutor Karim A.A. Khan KC, stated during a 2024 conference on cyber-enabled crimes held at ICC “We also need to be forward-looking, and ensure we are no less equipped to address the crimes of the future… Through Project Harmony, the Office is elevating its capacities to quickly review and disclose larger quantities of evidence left by crimes committed in modern-day conflicts”. Similarly, UN Secretary-General António Guterres highlighted at the AI for Good Global Summit, in Geneva 2019, that “We must seize the moment, in partnership, to deliver on the promise of technological advances and harness them for the common good" In addition to that, NGOs like Amnesty International and Human Rights Watch use AI-integrated tools to document evidence of human rights abuses in conflicts such as those in Ukraine and Syria.

Furthermore, Ukrainian law enforcement officials are beginning to use AI tools to document and investigate potential war crimes committed by Russian troops. More than 107,000 proceedings have been launched since the full-scale invasion, involving millions of documents and pieces of potential evidence. Yuriy Belousov, head of the war crimes department of the Office of the prosecutor general in Ukraine states "with the help of Microsoft tools, investigators can search by certain parameters and process huge amounts of information within a few hours."

Moreover, co-founder and executive director Roman Avramenko of Truth Hounds, a Ukrainian civil society organization which documents and investigates international crimes and human rights violations, explained that they are in the experimental phase of using a few advanced tools to streamline the documentation process. A significant challenge during this stage is establishing a coherent sequence of events, as witness testimonies often have inconsistencies when it comes to dates and details. To address this, they are leveraging AI technology to synthesize multiple testimonies gathered by different documenters in diverse locations.

Despite its benefits, the integration of AI into war crime investigations presents challenges. Biases in AI models can lead to biased conclusions. Distinguishing between genuine reports and disinformation is also difficult. Ethical concerns that arise involve the handling of sensitive data, confidentiality and testimonies. Moreover, errors and misinterpretations can occur due to over-reliance on AI. Thus, when integrating AI into war crime investigations it is essential to have robust human oversight to ensure accountability, transparency, and alignment with international legal standards.

In conclusion, AI represents a revolutionary tool that can transform war crime investigations. However, it is a double-edged sword. While AI-integrated weapons pose significant risks by removing human intervention from critical decisions, AI has the potential to bring perpetrators of atrocities to justice. With further research and ethical application, AI can become a powerful force for good—truly a blessing in disguise.

Keywords: #AI #Justice #WarCrimes #ICC