The U.S. government has lifted export controls on Anthropic's top AI models, Claude Fable5 and Mythos5. Competition is expected to intensify again among companies looking to deploy AI models in cyber security.

OpenAI has made a limited release of three new GPT-5.6 models: Sol, Terra and Luna. The top model, Sol, is specialised in cyber security. Chinese AI systems are also drawing attention after showcasing performance in some cyber security scenarios comparable to Anthropic's top model Mythos.

• U.S. government lifts export controls on Anthropic Fable5 and Mythos5 • OpenAI makes limited release of three GPT-5.6 models; top model specialised in cyber security • "Chinese AI has Anthropic Mythos-level power in security"; backlash to Trump's AI policy

Moves are also accelerating in the industry to apply agentic AI to security operations centres (SOCs). Logpresso signed an MOU with AI chip company FuriosaAI on an "NPU-based agentic SOC" and will advance its AI security monitoring technology. Network security company Fortinet launched FortiSOC, an integrated cloud-based SOC platform.

• Logpresso and FuriosaAI to cooperate on sovereign AI-based "agentic SOC" • Fortinet launches integrated SOC platform FortiSOC; deploys agentic AI

The article also summarised steps and issues involving companies in and outside South Korea around security.

Cyber security and cloud company Akamai will work with NVIDIA as part of efforts to introduce security architecture into "AI factories". The cooperation centres on embedding a zero trust architecture as a base layer into AI factories themselves to help companies protect data, context memory and autonomous agents.

• Akamai to cooperate with NVIDIA; embed zero trust security into "AI factories"

Amazon Web Services (AWS) highlighted three security strategy keywords as high-performance AI sharply reduces the time needed to find vulnerabilities and create attack code. They are Shift Left, continuous verification and AI agent security. Global open-source solutions company Red Hat will work with Palo Alto Networks and IBM to help companies discover vulnerabilities early and quickly apply protective measures across open-source software, commercial applications, operational technology (OT) and healthcare technology.

• "In the AI era, focus on Shift Left security; find and fix vulnerabilities during development" • Red Hat expands cooperation with IBM and Palo Alto Networks; supports corporate response to software vulnerabilities

AI agent security company Stryker has secured new investment worth $64 million. Stryker runs a platform that protects AI agents spreading across corporate systems.

• AI agent security startup Stryker raises $64 million

Hancomwith launched Hancom xFWeb, an installation-free security solution that protects web communications data from quantum computer hacking threats without requiring users to install a separate security program. Digital security company Initech announced a zero trust response strategy for the era of AI attacks. Initech said that in an N2SF transition environment, policies that apply differentiated access rights, usage scope, whether data can be taken out and log levels based on data grades such as confidential, sensitive and public are important. It also stressed that identity-based access control linking data importance and user permissions can be a key element for implementing N2SF. SKT participated in "Quantum Korea 2026" and unveiled advanced technologies and solutions under the theme of "next-generation quantum cryptography security for the AI and 6G era". Cryptocurrency exchange Bithumb will increase its bug bounty rewards to up to 200 million won. It said the move aims to block potential threats in advance by working with external security experts.

• Hancomwith launches Hancom xFWeb; supports PQC transition without system changes • Initech announces zero trust security strategy for AI era; actively responds to N2SF • SKT unveils next-generation quantum cryptography technology for AI and 6G era at "Quantum Korea" • Bithumb raises bug bounty rewards to up to 200 million won

The Korea Internet & Security Agency (KISA) will push ahead in earnest with new tasks for the safe use of personal information in the AI era. The Linux Foundation launched the "Acrites" initiative to efficiently address vulnerabilities in the open-source software ecosystem.

• KISA starts R&D on use of personal information, including deepfake suppression and privacy-by-design • Linux Foundation launches "Acrites"; joint response to open-source vulnerability patches

Frost & Sullivan said in its 2026 Cloud-Native Application Protection Platform (CNAPP) Frost Radar report that the focus of cloud security is shifting from simply securing visibility to reducing risks based on context.

• Frost & Sullivan: "Cloud security shifting from visibility to context-based risk management"

Palo Alto Networks and CrowdStrike rose 113 percent and 95 percent, respectively, from April to June, marking record quarterly gains. The rise was driven by a surge in corporate demand for cyber defence as sophisticated AI tools spread.

• Money pours in amid fear of AI hacking; Palo Alto up 113 percent, CrowdStrike up 95 percent