As AI spreads, security companies in South Korea and abroad are stepping up moves to expand AI-powered security products to respond to related threats.

CrowdStrike said at its annual conference that it added an identity framework for AI agents, parallel security investigations and a feature to preemptively block malicious open-source packages to its Falcon platform. It also unveiled SafeMind, an autonomous red-team system combining offensive AI and defensive AI. It also launched Falcon IQ, which automates vulnerability assessment and remediation tasks using more than 50 AI agents.

Cloudflare introduced Adaptive Intelligence, a persistent detection engine, to its Cloudflare Bot Management security platform that detects automated bot traffic and blocks malicious bots. Akamai Technologies is introducing Akamai Workforce Protector, a security layer, to its existing platform that protects applications, APIs and infrastructure.

ㆍCrowdStrike introduces an identity framework for AI agents… blocks malicious packages before execution ㆍCrowdStrike unveils SafeMind, an autonomous red team combining offensive and defensive AI ㆍCrowdStrike launches Falcon IQ… 50 AI agents work in place of security teams ㆍCloudflare launches Adaptive Intelligence... "It will sharply raise the cost of AI hacking" ㆍAkamai announces Workforce Protector, an AI security solution

South Korean security firm Pagonetworks unveiled a Security Operations strategy to respond to changes in the security environment in the AI era and also introduced 3 brands. ITCEN Group launched AgentGo Guard, an AI guardrail integrated security solution that supports enterprises and public institutions adopting generative AI and building agent ecosystems.

ㆍFrom detection and response to threat identification... Pagonetworks unveils AI-based security operations strategy ㆍITCEN Group launches AgentGo Guard, an AI guardrail integrated security solution

More cases are emerging of security products being developed using Nvidia open-source AI models. Capsule Security fine-tuned 2 Nvidia Nemotron models and released a detection system that determines AI agent behavior just before execution and allows, warns or blocks it in real time. CrowdStrike also developed an AI model specialized for cybersecurity with Nvidia.

ㆍCapsule Security unveils a system to prevent AI agent deviation based on Nvidia Nemotron ㆍCrowdStrike develops security-focused AI with Nvidia… also launches an AI lab

Market research firm Gartner also projected the AI security market will grow about 69 percent in 2027 from 2026 to about $4.783 billion. It forecast the market would expand to about $7.7 billion in 2028. Nikesh Arora (니케시 아로라), CEO of Palo Alto Networks, called for a wholesale overhaul, saying about $1 trillion worth of aging security infrastructure built for the pre-AI environment cannot handle automated threats.

ㆍ"AI security market to grow 69 percent next year"... Gartner forecast ㆍPalo Alto CEO "Existing cybersecurity cannot stop AI… $1 trillion infrastructure modernisation needed"

As the security industry rapidly reshapes around AI, mergers and acquisitions are also active among companies.

ㆍSecurity industry M&A accelerates amid AI spread... Palo Alto and Proofpoint pursue successive acquisitions

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

Naver Cloud was selected as the implementing agency for the government's "cybersecurity specialised AI foundation model" project.

ㆍGovernment selects Naver Cloud for "security-focused AI model" project ㆍNaver Cloud to develop a 700B-class "security foundation model"… pre-deploying 4,000 GPUs

TVING was confirmed to have suffered a breach that leaked information for 39.54 million accounts. Investigators found the attacker stole an access key held by a developer, infiltrated the internal development environment and then moved into the production environment to carry out a large-scale data leak.

ㆍTVING leaks 39.54 million accounts... includes dormant and withdrawn members ㆍTVING announces compensation plan for 40 million personal data leak... 5,000 points and subscription upgrade

The Model Evaluation and Threat Research Institute (METR) released results of an investigation into an incident in which OpenAI agents colluded with each other during testing and escaped onto the external internet. Visa said it is automatically handling everything from vulnerability detection to fixes and verification without human review through the open-source security tool VVAH (Visa Vulnerability Agentic Harness).

ㆍInvestigating AI hacking with AI finds it sides with the attacker ㆍVisa releases VVAH as open source... automates detection, fixes and verification

IT services company HNIX spun off its information security business unit and set up security specialist company HNSLAB. JiranJigyo Security signed a memorandum of understanding with cybersecurity startup Pilsang on cooperation in mobile security technology and business. FasooAI launched DRM Source Code Security, a source code security solution using a virtualised isolation approach.

ㆍHNIX security unit reborn as security specialist company HNSLAB ㆍJiranJigyo Security cooperates with Pilsang... strengthens AI-based mobile threat detection technology ㆍFasooAI launches virtualised source code security solution