As competition in the tech industry is reorganising around AI agents, the balance of power in the security sector is also shifting. More cases are emerging in which established tech companies strengthen AI agents and extend into security. Snowflake, which grew as a data platform, is also expanding into AI and accelerating efforts to bolster security capabilities.

• Snowflake gains momentum in security expansion..."Agentic AI security needs a fundamentally different approach" • Snowflake upgrades Horizon Catalog...integrates governance, business context and security • From coding to security...Snowflake-Anthropic expand enterprise AI alliance

Mayank Upadhyay (마얀크 우파댜이), Snowflake's chief trust and security officer, said all security companies are reinventing themselves for the AI era. He said AI is moving so fast that keeping up has become a core challenge, and new use cases that were unimaginable keep emerging. He said agents will become smarter, making visibility and governance for agents deployed inside companies more important. He said that for threats attacking from outside using AI, automated source code scanning and automated response are key. Ultimately, he said, AI must collect and analyse data to find and fix problems, because people alone cannot keep up with AI speed.

• "Security must be close to data...Snowflake's role will grow"

Microsoft unveiled MDASH (Microsoft Security Multimodal Agentic Scanning Harness), a platform that uses AI agents to automatically detect software vulnerabilities. Microsoft also added a feature to its endpoint security platform Defender for Endpoint that automatically isolates infected devices from the network.

• Microsoft unveils AI vulnerability scanning platform MDASH...uses more than 100 AI models • Microsoft adds automatic isolation of infected devices to endpoint security product

Workday, a cloud-based enterprise software company, also introduced Agent Passport, a feature that independently verifies whether security and compliance standards are met before an agent is deployed. Cisco also strengthened security functions for AI agents. "Defenseclo" is a security and governance framework for local AI agents such as OpenAI Codex, Claude Code and OpenClaw. It checks vulnerabilities and applies access controls, and supports monitoring and governance by linking with Cisco AI Defense and Splunk. Cisco also introduced zero-trust expansion features that more finely control AI agent behaviour and access resources, and "Agentic SOC", which automates security operations centre tasks.

• Workday upgrades AI agent development platform...adds developer agents and AI security • Cisco unveils 'Cloud Control'...integrates AI agent operations and security

Anthropic added 150 organisations that can access Mithos, which it has disclosed only on a limited basis. The list includes South Korean companies and institutions.

• Anthropic expands Mithos access to 150 new institutions in 15 countries..."includes Samsung, SK Hynix and SKT" • SKT joins Anthropic 'Glasswing'..."contributing to strengthening digital security"

Discussion around Mithos has so far focused on what vulnerabilities were found and how quickly. Whether the process incurred costs was largely unclear, but a recent report by The Information suggests that for companies without deep pockets, even if they can use Mithos, using it may not be easy.

• [Tech Inside] Cybersecurity economics changing in the wake of Mithos

In South Korea, industry moves toward the quantum security market are accelerating.

Quantum technology company SDT signed a business partnership with video security company Viva, centred on joint development and mass production cooperation for "QRNG (quantum random number generation) encrypted AI CCTV". Korea Quantum Computing (KQC) completed a proof of concept for quantum-resistant cryptography (PQC)-based security technology to prepare for hacking using agentic AI and risks from quantum computers, with LS ITC, the IT services company of LS Group. RaonSecure supplied a PQC solution to KDB Life Insurance.

• SDT-Viva cooperate to develop 'QRNG quantum security AI CCTV' • Korea Quantum Computing completes PQC-based proof of concept with LS ITC for authentication, access and certificate automation • RaonSecure supplies PQC solution to KDB Life Insurance

A personal information leak occurred at South Korean online video service (OTT) platform Tving, and concerns about secondary 피해 are spreading as connected information (CI) is also included among the leaked items.

• 'Digital resident registration number' also stolen in Tving hacking...concerns over secondary damage spread