▶ DeepBrain AI enhances its solution by adding 2 million data samples to the NIA deepfake data construction project.
▶ The service supports detection of videos, images, and audio, analyzing them within 5 to 10 minutes to determine authenticity, alteration rates, and synthesis types.
▶ It holds the largest number of deepfake detection references in the country, collaborating actively with public institutions, including supplying the National Police Agency.
.png)
DeepBrain AI, a leading generative AI company in South Korea, announced that it will provide a one-month free trial of its ‘Deepfake Detection Solution’ to companies, public institutions, and educational organizations in need of deepfake detection.
According to the National Police Agency, crimes related to the creation of false video content using AI have surged over the past four years. In 2021, there were 156 reported cases, which rose to 180 last year, and the first half of this year alone has already seen 297 cases, reflecting a rapid increase. Notably, the distribution of deepfake pornography, which involves synthesizing the faces of acquaintances, has spread widely across schools, organizations, and military units, causing significant social repercussions.
To minimize the damage caused by deepfake-related crimes and facilitate smooth investigations and responses, DeepBrain AI has decided to offer its deepfake detection solution free of charge for one month to educational institutions, public agencies, B2B companies, and enterprises.
DeepBrain AI’s deepfake detection solution allows users to upload suspected deepfake videos to the system, which then instantly determines their authenticity and provides results. The company has successfully completed a deepfake data construction project with the Korea Information Society Agency (NIA), enhancing the solution. In addition to the previous training of 5.2 million samples, it has conducted additional training with 2 million data samples, totaling 2,000 hours of high-quality data, improving the overall quality of the dataset.
The deepfake detection solution consists of comprehensive detection that supports video and image detection, as well as voice detection. Results are categorized as real (Real) or fake (Fake), and data on alteration rates and synthesis types are also provided. It can detect various types including face swap, lip sync, and generative video, with a detection time of approximately 5 to 10 minutes.
DeepBrain AI provides the deepfake detection solution in two formats: on-premises and Software as a Service (SaaS). The SaaS version can be easily accessed from anywhere with an internet connection. For more details, inquiries can be made through the official DeepBrain AI website.
DeepBrain AI holds the largest number of deepfake detection references in South Korea. Recognized for its excellent detection capabilities, it supplied the deepfake detection solution to the National Police Agency in March of this year and is actively collaborating with various public institutions on deepfake detection projects and related research.
CEO Jang Se-young stated, “As deepfake crimes that exploit AI technology continue to escalate, we have decided to offer our deepfake detection solution for free to businesses and public institutions that require authenticity verification in order to minimize related damages.” He added, “We will continue to expand our comprehensive support to effectively respond to deepfake crimes through collaboration with various organizations and companies.”
Meanwhile, DeepBrain AI, as a specialist in generative AI and deepfake detection, is expanding its AI human business across various sectors, including finance, media, education, commerce, and public service, both domestically and internationally. The company has collaborated with high-profile individuals such as President Yoon Suk-yeol, news anchor Kim Joo-ha, and professional golfer Choi Kyung-ju, establishing a prominent position in the AI human field. Its accumulated investment has reached 60 billion won, and it is preparing for an IPO next year.
