![[AI Pick] People are looking for AI errors and risks… Launch of ‘AI Safety Guardian’](/api/img?u=https%3A%2F%2Fimg.yna.co.kr%2Fphoto%2Fcms%2F2026%2F10%2F26%2F06%2FPCM20261026000106990_P2.jpg&w=1200&q=72&f=webp)
[AI Pick] People are looking for AI errors and risks… Launch of ‘AI Safety Guardian’
Quick Look
The Ministry of Science and ICT and the Artificial Intelligence Safety Research Institute have appointed 80 'AI Safety Guard' monitoring groups, where the public directly reports errors and risk cases of generative AI, and will operate them until August next year.
AI-generated summary
Why It Matters
The Ministry of Science and ICT has been conducting a citizen-participatory 'Strong AI Challenge without Information Errors' campaign for about 30 days since December last year, and achieved the result of confirming 464 out of a total of 625 reports as valid cases of errors, and formed a monitoring group to make this regular.
A constant monitoring system is operated in which the public directly finds and reports incorrect information, biased answers, and security vulnerabilities generated by artificial intelligence (AI).
The Ministry of Science and ICT, along with the Artificial Intelligence Safety Research Institute, held an inauguration ceremony for the citizen-participatory AI monitoring group 'AI Safety Guardian' at the Glad Hotel in Yeouido, Seoul on the 12th.
The AI Safety Guard is a monitoring group that discovers and reports errors and risk cases in generative AI. Of the 138 people who applied for two open recruitments in August and September, 80 were selected and will work until August of next year.
The government conducted a citizen-participatory 'Strong AI Challenge without Information Errors' campaign for about 30 days starting in December last year, and achieved the result of confirming 464 out of a total of 625 reports as valid error cases, and formed a monitoring group to make this regular.
Cases reported by the monitoring group are confirmed after verifying their effectiveness based on accuracy, risk, reproducibility, and usability.
Confirmed cases will be used as basic data for AI safety evaluation and risk research, and will also be shared with related AI companies.
Open Questions
- When will the results of the monitoring group's activities be made public?
- What follow-up actions will be taken after confirmed cases are shared with AI companies?
![[TechSnap] AI from marketing planning to execution… Enhance ‘Mopai’ released](/api/img?u=https%3A%2F%2Fimg.yna.co.kr%2Fetc%2Finner%2FKR%2F2026%2F11%2F12%2FAKR20261112066400017_01_i_P2.jpg&w=320&q=72&f=webp)
