[Edaily Reporter Ahn Yu-ri] With Anthropic unveiling its new artificial intelligence (AI) models, "Mitos 5" and "Fable 5," a major shift in the domestic security market is expected to begin in earnest.

As vulnerability analysis and penetration testing—tasks previously performed primarily by experts—become feasible for AI, small-scale security firms specializing in these services are expected to be directly impacted.

On the 9th (local time), Anthropic unveiled "Mitos 5," touted as the most advanced AI model currently available, and "Fable 5," a version based on Mitos 5 that includes safety features making it accessible to general users.

Mitos 5 and Fable 5 are evaluated as demonstrating not only excellent coding and task-handling capabilities but also outstanding performance in the field of cybersecurity.

In the "ExploitBench" evaluation, which measures cybersecurity capabilities, Mythos 5 scored 78%. This outperformed not only OpenAI’s GPT-5.5, which scored 34%, but also the previously released Mythos preview (69%) from last month.

The company emphasized that Fable 5 boasts the highest level of performance among existing open-source AI models.

Industry observers predict that with the launch of Fable 5, tasks such as vulnerability analysis, attack scenario verification, and penetration testing—areas previously handled by skilled white-hat hackers and specialized personnel—will be largely automated.

Kim Hyuk-jun, CEO of Naru Security, stated, “Vulnerability analysis has traditionally been a task that could only be performed by humans,” adding, “It is highly likely that companies with a technological edge will dominate the market going forward.”

He continued, “Just as the operating system market eventually consolidated around a few major companies, the security market could follow a similar path,” predicting that “simple vulnerability analysis tasks previously performed by small companies could largely disappear.”

According to the 2025 Information Security Industry Survey, 794 out of 876 domestic information security companies—90.7%—are small and medium-sized enterprises (SMEs). Of these, 47.5% are small businesses with annual sales of 5 billion won or less.

However, the industry anticipates that rather than the demand for vulnerability analysis disappearing entirely, the market structure will shift. As hacking threats driven by AI increase, demand for cybersecurity is rising, and there is a forecast that the intermediate response market—positioned between prevention and post-incident response—will experience new growth.

Domestic security companies are also taking steps to respond to the growing AI-driven threats and the tectonic shifts in the rapidly changing cybersecurity market. Recognizing that traditional security solutions alone are insufficient to counter AI-based attacks, they are accelerating the development of next-generation security products utilizing Large Language Models (LLMs) and AI.

Following the launch of its first product line for the AI-based vulnerability analysis solution “AlienRay” last month, Steelean released “WebSuit,” a web tampering prevention and protection solution incorporating AI-based behavioral analysis, this month.

#RaonSecure is currently developing a total of five solutions, including an agentic AI-based security automation platform, in collaboration with AI security provider Upstage. The company aims to launch AAM (Agentic AI Management) within the year, which will enhance the reliability of the entire system by controlling the permissions under which AI operates.

A RaonSecure representative stated, “AAM is currently under development with a target launch date within this year, and the AI Security Officer (solution), where Agentic AI acts as a security officer, has reached the MVP (minimum viable product) stage and is currently undergoing verification.”