Ruling and Opposition Lawmakers Clash Over Sovereign AI Foundation Model at National Assembly Audit; People Power Party Counters with 'AI Daejang-dong' Claims

Ruling and opposition lawmakers crossed swords on the 6th over the government's sovereign AI foundation model development project during a National Assembly audit of the Ministry of Science and ICT held by the Science, ICT, Broadcasting and Communications Committee. The People Power Party denounced the project as an 'AI Daejang-dong,' questioning its effectiveness and fairness, while the Democratic Party countered by stressing its necessity, calling it 'forward-looking.'

Kim Jang-kyeom of the People Power Party raised concerns about the fairness of the evaluation process. He pointed out that the dataset lead contractor has annual revenue of 150 million won with eight employees, while the knowledge lead contractor has annual revenue of less than 1 billion won with fewer than 20 employees. Given that the evaluation targets are Korea's top-tier AI companies, he likened the process to "high school students setting the questions for a graduate school exam."

The Democratic Party pushed back, defending the project's necessity. Han Joon-ho, the ruling party's committee secretary, said the sovereign AI foundation model and the frontier AI project are separate initiatives, arguing that technology and human resources should be leveraged while preventing overlap in GPUs and budgets. Rep. Lee Joo-hee also praised the project, saying, "Our government was remarkably forward-looking." Kim Hyun of the Democratic Party demanded that the opposition stop disparaging the Lee Jae-myung administration's policies, asking, "If it weren't for the sovereign AI foundation model and the AI Basic Act, where would South Korea stand in the global AI race?"

Bae Kyung-hoon, Deputy Prime Minister and Minister of Science and ICT, who attended the audit, said in his work report that the sovereign AI foundation model project would proceed without wavering, though duplicate investments would be prevented. He also announced plans to replace the model currently used for the government-wide shared AI infrastructure with the second-generation sovereign AI foundation model.