The Gangnam Police Station submitted the disputed audio files to the National Forensic Service (NFS) for detailed examination. After months of rigorous testing, the NFS concluded that while it could not definitively prove manipulation in every respect due to the absence of original source files, clear technical indicators pointed to heavy post-production editing. More importantly, police have now disclosed that they employed advanced forensic software specifically designed to detect AI-generated or AI-assisted audio. These tools include spectrogram analysis, voice synthesis artefact detection, waveform consistency checks, and emotional prosody evaluation algorithms.
The software identified several tell-tale signs commonly associated with AI voice cloning technology, such as unnatural frequency patterns, artificial background noise, and micro-irregularities in speech rhythm that are virtually impossible to replicate manually. Tools similar to those used by international forensic laboratories – capable of distinguishing genuine human recordings from deepfake audio – revealed inconsistencies that strongly suggest the files had been processed or partially generated using AI software.
The recordings in question were made public during a controversial press conference organised by the YouTube channel Garosero Research Institute in May 2025. In the clips, a voice purportedly belonging to Kim Sae Ron claimed she had been in a relationship with Kim Soo Hyun since 2019, when she was still a minor, and alleged financial pressure from his side. Kim Soo Hyun immediately denied the claims, describing them as fabricated, and filed defamation lawsuits against those involved.
The latest forensic findings have dramatically shifted the legal landscape. Prosecutors have now requested an arrest warrant for Kim Se Ui, head of Garosero, on charges including spreading false information and defamation. A pre-trial detention hearing is scheduled for 26 May 2026. If proven that AI was deliberately used to create misleading evidence, the case could result in severe criminal penalties and set an important precedent regarding the misuse of artificial intelligence in public accusations.
This development highlights the growing sophistication of both AI voice generation tools – such as Respeecher, ElevenLabs, or domestic Korean equivalents – and the forensic technology designed to counter them. Experts note that modern deepfake audio can sound remarkably authentic to the untrained ear, making cases like this particularly challenging. The police’s use of multiple cross-verification methods demonstrates how seriously authorities are now treating digital evidence in high-profile disputes.
Kim Soo Hyun’s legal team has welcomed the findings, describing them as validation of their long-standing position. Meanwhile, representatives for Kim Sae Ron’s family and Garosero maintain that the recordings reflect genuine conversations, arguing that an inconclusive result does not equate to proof of fabrication.
The scandal has had far-reaching consequences, severely damaging reputations, careers, and public trust. It has also sparked intense national debate about ethics in online journalism, the responsibilities of YouTube content creators, and the urgent need for updated legislation to address AI-generated content. Many observers warn that deepfake technology is becoming a dangerous weapon in personal and professional disputes.
As the investigation continues, police are examining digital devices and interviewing key individuals to trace the origin of the audio files. While the final verdict is still pending, the disclosure that specialist AI-detection software was used represents a critical breakthrough. It underscores a broader truth in the digital age: audio evidence can no longer be taken at face value. Advanced technology can both create convincing falsehoods and expose them with increasing precision.
Regardless of the ultimate outcome, this case has already exacted a heavy toll on all parties involved. It serves as a sobering reminder of the power – and peril – of artificial intelligence in shaping public narratives and personal destinies.
