by Lee Myeonghwan
Published 25 Sep.2025 09:05(KST)
SK Telecom's artificial intelligence (AI) service A.Dot is introducing the latest large language models (LLMs), including OpenAI's 'GPT-5.' A new feature has also been added to alert users if received text messages are identified as spam or potentially dangerous.
On September 25, SK Telecom announced that it has upgraded A.Dot by integrating the latest global AI LLMs and adding a new text message security feature.
First, SK Telecom has added the latest LLMs to A.Dot's 'Search' service. The newly introduced AI models include the latest version of SK Telecom's proprietary LLM, A.DotX 4.0, as well as OpenAI's latest high-performance AI models from the GPT-5 series (GPT-5, GPT-5 Mini, GPT-5 Nano).
According to SK Telecom, A.DotX 4.0 has received high marks for its understanding of the Korean language and Korean culture. Based on evaluation results published in July this year, A.DotX 4.0 scored 78.3 on the KMMLU, a Korean language proficiency benchmark, surpassing GPT-4o's score of 72.5. It also achieved a score of 83.5 on CLIcK, a benchmark for Korean language and culture, outperforming GPT-4o's 80.2.
Three models from OpenAI's latest high-performance GPT-5 series have also been added. The GPT-5 series offers enhanced capabilities in handling complex tasks and reasoning compared to previous models, while also reducing the likelihood of hallucinations. Users can now access a total of 19 advanced AI models, including the newly added models as well as Claude, Perplexity, Gemini, and Liner.
The AI phone service 'A.Dot Phone' has also introduced a text message security feature called 'AI Message.' AI Message uses AI technology to label incoming texts suspected of being spam or phishing with tags such as impersonation, scam, or link warning, thereby alerting users to potential risks.
The AI model determines whether a message is suspected spam or phishing by comprehensively analyzing factors such as the sender's number, message content, and included links (URLs) based on spam and phishing-related data accumulated in A.Dot Phone. The 'impersonation warning' label is applied to messages impersonating trusted entities like family members or financial institutions. The 'scam warning' label is for messages that attempt to lure recipients with offers such as loans or job opportunities. The 'link warning' label is used for messages containing links that may lead to malicious app installations or personal information theft.
If a user clicks on a message labeled with 'link warning' due to the presence of a risky link, A.Dot Phone displays a warning pop-up before redirecting to the default messaging app, alerting the user again to the potential danger. A 'hide cautionary messages' feature has also been added, allowing users to configure the app so that messages requiring caution are not displayed at all.
To enhance call security, SK Telecom has previously offered features such as warnings for suspected spam or voice phishing numbers, sharing of user evaluation information, and 'AI Safe Block,' which automatically blocks calls from suspicious numbers using AI technology.
AI Message is available on the A.Dot Phone app for Android operating systems and in the 'Phone' menu within the A.Dot app for iOS. It is not supported in the default messaging app. The AI Message feature can be activated either through the pop-up that appears after enabling the 'View Messages Together' option in the 'Recent Records' tab (Android) or via the 'AI Message' menu in settings (iOS). Users must agree to the terms and conditions to use the service.
Kim Yonghun, Head of SK Telecom's A.Dot Business Division, stated, "By applying the latest AI models to A.Dot and introducing new text message security features, we have enabled customers to use the latest AI models more conveniently while ensuring safe service based on AI technology. We will continue to enhance A.Dot so it can be a reliable AI companion that safeguards our customers' daily lives."
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.