SK Telecom said on Feb. 25 it will participate in the Global AI Red Team Challenge held during MWC26 to verify the safety and reliability of its self-developed artificial intelligence (AI) model, A.X K1.

The challenge will be conducted by having about 100 evaluators identify misuse risks such as generating discriminatory responses and information breaches in large language models (LLMs) developed by telecom and AI companies in and outside South Korea.

The event is jointly hosted by the GSMA and Selectstar, an AI data and reliability evaluation specialist. It will be held on March 3 local time at the Talent Arena inside the MWC venue in Barcelona, Spain. Verification results derived by the evaluators will be converted into scores through a panel of judges. The number of participating companies and final scores and rankings will not be disclosed.

SK Telecom will take part in the challenge with A.X K1, which it developed independently across the entire process from model design to training. A.X K1 was selected in the first-stage evaluation of the national AI project, the independently developed AI foundation model. The A.X model series is an AI model that SKT has advanced based on data accumulated through its telecommunications and platform operations. It is currently applied to AI services such as A.Dot and is being expanded for internal 업무 within the SK Group.

Tae-yoon Kim (김태윤), head of SKT foundation models, said, "As AI spreads widely across customers' daily lives and society as a whole, securing safety and reliability is becoming more important along with technological competitiveness." He added, "Through participation in the Global AI Red Team Challenge, we will make it an opportunity to further enhance trust in SKT's AI technology."