As the global security market rapidly reshapes around AI, Korean security companies are also moving faster on AI.

East Security, the security unit of AI services company ESTsoft, is launching Alyac Agentic EDR, an endpoint detection and response solution linked to LG AI Research’s K-EXAONE AI model introduced for a government-led, homegrown foundation model project.

East Security also signed an MOU with AI chip company FuriosaAI to develop next-generation K-AI security solutions and cooperate on technology. Offensive cybersecurity company Theori supplied its AI-based black-box penetration testing solution Xint Web to Woori Financial Group. Cloud and AI managed services company Cluocus is also accelerating its AI-based security business. It recently attended a domestic security event with global cloud and AI security company Wiz, examining key security challenges companies face as cloud and AI environments spread rapidly and presenting practical response measures to address them.

East Security unveils Alyac Agentic EDR linked to LG’s K-EXAONE AI model. East Security and FuriosaAI cooperate to develop AI security infrastructure. Theori supplies AI penetration testing solution Xint Web to Woori Financial Group. Cluocus and Wiz take part in ISEC 2026 and present corporate security strategies in the AI era. Cluocus partners with global company Stryker to target the agentic AI security market. Cluocus partners with global security company Panzer to expand its AI SOC platform in South Korea.

M&A deals are also emerging. Nets, an IAM and SSO specialist, completed the acquisition of API security specialist Softfreak. Nets will combine Softfreak’s API security, access control and AI capabilities with its existing IAM and SSO business to accelerate the buildout of a next-generation identity security platform.

Nets acquires management control of Softfreak and expands integrated identity security.

A hacking incident involving Hugging Face has put companies on alert across their security systems as AI agents emerge as actual cyberattack actors. The security industry sees the incident not as an exception but as a signal that attacks exploiting autonomous AI are set to accelerate, and it is speeding up efforts to prepare countermeasures. About 120 organisations including Nvidia, Cisco and CrowdStrike are preparing a joint framework to systematically report and record AI agent security incidents. Microsoft said AI diffusion has broken the scarcity of cyberattack capabilities and stressed that security responses should shift from post-incident patching to safer software development.

A string of incidents at OpenAI, Anthropic and Meta puts corporate responses to AI hacking threats to the test. About 120 organisations including Nvidia and Cisco push a joint AI agent incident reporting system. "In the AI era, fast patching has limits.... software must be developed safely."

The article also summarised moves and issues involving companies in South Korea and abroad around security.

OpenAI expanded and reorganised its cybersecurity programme Daybreak into two tracks, Daybreak Blue and Daybreak Red, and will provide Red users with a new model called GPT-5.6-Cyber. OpenAI also halted some work related to its planned AI model Astra. It cited internal evaluation results, saying it could not rule out the possibility that the model has cyberattack capabilities that would qualify for a "critical" rating.

OpenAI expands Daybreak security programme and unveils a security-focused model. OpenAI halts some Astra model development, saying it cannot rule out "serious cyberattack" potential.

Cybersecurity company Palo Alto Networks applied OpenAI’s most advanced cyber model inside customer environments and expanded its service to analyse attack paths.

Palo Alto Networks deploys OpenAI inside customer networks and analyses real attack paths.

Data security company Hanssak signed a master distribution agreement for an integrated security solution with information security company SMPeoples. AI security company StealCut, which blocks and detects deepfake generation, secured seed funding from Mashup Ventures.

Hanssak signs distribution deal with SMPeoples to expand distribution for its flagship security solutions. Deepfake-blocking AI security startup StealCut secures seed investment from Mashup Ventures.

Israeli cybersecurity AI startup Korma raised $60 million in seed funding to develop defensive cybersecurity AI.

Korma raises $60 million and bets on an AI model specialised in defending against cyber threats.

The National Office of Investigation at South Korea’s National Police Agency distributed a joint South Korea-U.S. cybersecurity advisory on the GUNRA ransomware with the U.S. Federal Bureau of Investigation, the Cybersecurity and Infrastructure Security Agency, the National Security Agency, the Defense Cyber Crime Center and the U.S. Secret Service.

South Korea police warn of the spread of GUNRA ransomware and distribute a joint South Korea-U.S. cybersecurity advisory.

It was later confirmed that a Claude-based AI agent exploited an authorisation verification vulnerability in a gym booking system and cancelled other users’ reservations.

A gym booking AI agent cancels other people’s reservations and cuts in line.

The U.S. government for the first time allowed some private companies to take part in offensive cyber operations targeting international criminal organisations and hackers. The White House said in a presidential message it would use private sector capabilities to respond to cyber threats targeting Americans, including ransomware, financial fraud and sexual extortion blackmail.

United States allows private companies to participate in offensive cyber operations for the first time.