[Click eStock] "CrowdWorks Expected to Benefit from Deepfake Prevention"

Independent research firm ValueFinder analyzed on the 11th that data labeling by CrowdWorks is necessary to prevent deepfakes.

Lee Chungheon, a researcher at ValueFinder, explained, "We won the project to build AI training data for deepfake prevention videos and landmark images promoted by the Ministry of Science and ICT in 2020," adding, "We hold a unique position in developing deepfake solutions."

He continued, "Recently, deepfake crimes have surged, becoming a social issue," and added, "Detecting deepfakes requires 'real data and fake data'." The researcher judged that the importance of companies like CrowdWorks, which supply training data, will increase.

CrowdWorks operates the largest data labeling platform in Korea. It provides data necessary for AI training. It has experience building AI training data for deepfake detection in cooperation with the National Information Society Agency. It operates a 'Data Engine' that preprocesses AI training data, 'LLM' that builds customized AI models, and 'Crowd Academy' that trains data labelers.

Its sales in the first half of this year decreased by 64.7% compared to the previous year. The researcher analyzed, "Considering the delivery schedule of AI construction projects, related sales recognition will be possible from the second half of this year." CrowdWorks' order backlog in the first half was 8 billion KRW.

The researcher forecasted, "CrowdWorks invested shares in the smart factory specialized company 'Sizzle' and will accelerate the AI solution business for small and medium manufacturers and local data education business in Japan through a capital increase of its Japanese subsidiary."