LG Uplus (032640) announced on the 19th that it will participate in the Global AI Red Team Challenge held during Mobile World Congress to verify the safety and reliability of generative artificial intelligence.
The Global AI Red Team Challenge is an AI verification program organized by the GSMA and will be held in Barcelona, Spain, where MWC26 takes place. LG Uplus will submit its proprietary telecom-specialized AI model "ixi-GEN" to test its resilience against various risk scenarios that could occur from actual user perspectives.
Approximately 100 participants will join the challenge as "red team" members, tasked with finding AI vulnerabilities solely through prompt design. The program features conversation and prompt-based attacks rather than technical hacking to verify whether AI policy and safety mechanisms can be bypassed. The evaluation criteria focus on whether the AI produced responses it should not have, rather than response accuracy. This approach aims to examine both indirect and potential risks that generative AI could pose.
Through this challenge, LG Uplus plans to objectively verify ixi-GEN's safety and reliability and continuously enhance its AI model based on identified vulnerabilities. The company also intends to expand discussions on AI security and responsible utilization through exchanges with global telecom carriers and AI experts.
"As AI is rapidly being applied across telecommunications and customer services, verifying reliability and safety is more important than ever, not just performance," said Lee Hye-jin, Head of Technology Strategy at LG Uplus. "Through participation in the Global AI Red Team Challenge, we will further strengthen ixi-GEN's competitiveness and provide AI services that customers can use with confidence."