[Exclusive] Prosecutors Indict Case Using AI Analysis of KakaoTalk Messages for Trade Secret Leak

In a groundbreaking application of artificial intelligence in the judicial system, South Korean prosecutors have successfully indicted three individuals accused of leaking trade secrets and committing breach of trust. The Seoul Eastern District Prosecutors' Office employed advanced AI techniques to sift through an enormous volume of KakaoTalk messages, a feat that proved instrumental in uncovering crucial evidence that had eluded traditional investigative methods.

“

Artificial intelligence can fabricate or fail to find content if it is not sufficiently trained on the data, so we used a method of refining the KakaoTalk records into a database for the AI to learn.

The case involves developers accused of illicitly copying and transferring the source code of a proprietary overseas shopping mall purchasing program to a rival company. The complexity of the case was amplified by the technical jargon prevalent in the defendants' communications and the sheer volume of data, making manual analysis an arduous, if not impossible, task. Prosecutors initially struggled to piece together the criminal narrative based solely on the victim's testimony and keyword searches.

“

We used artificial intelligence to find clues as a cross-verification measure since it is impossible to go through the vast conversations one by one.

However, the integration of AI into the investigative process marked a turning point. By training the AI on the defendants' KakaoTalk conversations and the program code itself, investigators were able to extract damning evidence that highlighted the intent and actions related to code replication. The AI not only identified direct mentions but also inferred criminal intent from nuanced conversations, such as phrases indicating a desire to mimic or closely replicate competitors' programs, even without using explicit terms like 'copying.' This capability allowed prosecutors to build a stronger case, moving beyond simple keyword matches to understand the context and implications of the digital communications.

“

We were able to find evidence that was not found through keyword searches conducted during the police investigation stage.

This case represents a significant advancement in how law enforcement can leverage AI for complex investigations. As prosecutor Son Sung-min explained, the AI was used as a cross-referencing tool to find leads within the vast dataset, requiring careful data refinement to ensure the AI's effectiveness. The successful indictment demonstrates the potential of AI to enhance investigative efficiency and accuracy, particularly in cases involving digital evidence and sophisticated trade secret theft, offering a glimpse into the future of legal proceedings in the digital age.

“

We were able to secure conversations containing evidence of code duplication by finding parts that seemed suspicious of criminal activity based on the context of the conversations, even if words like 'copy' were not directly included.