“Sharing AI Model Vulnerabilities Secured Through International Cooperation”
Baek Gung-hoon, Deputy Prime Minister and Minister of Science and ICT, attended the Science and Technology Ministers’ Meeting held on the Government Complex Seoul in Jongno-gu, Seoul on the thirtieth, talking on present points such because the coverage route for knowledge promotion within the period of AI transformation. 2026.4.30 Photo by Jo Yongjun
The authorities is shifting to set up a joint public-private response system to counter evolving synthetic intelligence (AI) cyber threats. It will centralize the administration of AI vulnerabilities and patch data whereas strengthening technical assist for companies and authorities companies. In addition, beginning subsequent 12 months, the home data safety system will likely be transitioned to an unbiased AI technology-based system to put together for the routine safety utilization of high-performance AI.
On May 29, the Ministry of Science and ICT introduced its “Private Sector Information Security Promotion Plan to Respond to AI-based Cyber Threats” on the ninth Science and Technology Ministers’ Meeting.
Recently, main U.S. huge tech firms have launched initiatives to present AI fashions with cybersecurity capabilities on the stage of safety specialists completely to choose firms. According to the primary report of Anthropic’s Glasswing Project in May, greater than 16,000 vulnerabilities had been discovered within the undertaking members’ software program and open-source property.
As an preliminary step, the federal government will set up a joint public-private response system to tackle the general public disclosure of AI vulnerabilities. Centered on the National Security Office of the Blue House, the federal government will make sure the speedy sharing and dissemination of data on AI vulnerability disclosures, patches, and menace conditions, and can arrange an emergency system for joint response within the occasion of safety incidents. The Ministry of Science and ICT will function a central state of affairs room, whereas every related ministry will run its personal state of affairs room for the non-public sector.
A Vulnerability Management Center will likely be arrange inside the Korea Internet & Security Agency (KISA) to centralize the administration of vulnerabilities and patches, and to promote technical assist for related ministries and companies. The KISA Vulnerability Information Portal (KNVD) would be the hub for in depth assortment and evaluation of vulnerabilities and patches via public disclosure, experiences, and sharing with associated companies. This data will likely be quickly shared with roughly 28,000 Chief Information Security Officers (CISOs) at firms, non-public cooperation channels (C-TAS, ISAC), and all authorities and navy state of affairs rooms, forming an emergency response system that swiftly points motion suggestions.
The Ministry of Science and ICT additionally plans to pilot the usage of the most recent high-performance AI fashions—secured via worldwide cooperation—for work associated to vulnerabilities and patches, in addition to for general enterprise assist.
For main firms that might endure vital injury from AI safety threats, safety preparedness will likely be strengthened. Under the supervision of every accountable ministry, these firms will likely be required to independently handle asset administration, vulnerability inspections, and patch responses, whereas the federal government will conduct sector-specific implementation checks.
For small and medium-sized enterprises (SMEs), the federal government will distribute an online software that diagnoses present safety ranges and recommends safety funding pointers and motion steps accordingly. It may even assist know-how for producing and analyzing software program payments of supplies (SBOMs) to proactively determine and tackle open-source vulnerabilities which are simply exploited by AI.
The Ministry of Science and ICT plans to present infrastructure, using high-performance AI fashions to which it has entry, for vulnerability inspections and different assist for SME software program.
Additionally, to counter AI safety threats, the federal government will repeatedly monitor domains worldwide and detect and reply to AI-based malicious actions and domains as quickly as they’re created. In the occasion of incidents or suspected incidents involving AI providers, the “Incident Investigation and Review Committee” will likely be instantly convened to guarantee a swift investigation and stop the unfold of injury.
The Ministry of Science and ICT will start by securing OpenAI’s Government Trust-based Access Program (GTAC) and can proceed cooperation to take part in AI safety initiatives with international huge tech firms. The ministry may even strengthen collaboration with cybersecurity companies in allied nations for AI-based menace response.
Looking forward, the federal government plans to absolutely transition the home data safety system to an unbiased AI technology-based mannequin from subsequent 12 months, in preparation for the on a regular basis safety use and potential weaponization of high-performance AI. Various initiatives will likely be deliberate and executed to set up AI safety sovereignty.
Hot Picks Today
“Volunteers Line Up to Serve: 5,000 Apply for Military Service in France, with 20% Women”
Baek Gung-hoon, Deputy Prime Minister and Minister of Science and ICT, acknowledged, “The pace of AI development in the field of cybersecurity is now comparable to that of top-level hackers. If Korea fails to establish a security system and global collaborations suitable for the AI era, its ambitions to become a global top-three AI power could be shaken. Through these measures, we will establish an emergency response system to address large-scale AI-related vulnerability disclosures and will swiftly pursue the establishment of AI security sovereignty based on our own technologies and models.”
This content material was produced with the help of AI translation providers.
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.