[Digital Today reporter Chi-gyu Hwang (황치규)] As AI is rapidly being used in cybersecurity, security patching practices across the tech industry are also shifting. In terms of the speed and scale of patch releases, different scenes from the past are unfolding in many places.

Microsoft has recently released its biggest-ever security patch, helped by AI-based detection of code vulnerabilities. The company explained that internal staff are using AI to find more software security flaws that had not been discovered before.

· Microsoft uses AI to patch the most security vulnerabilities ever

Moves by governments in various countries to speed up vulnerability patching are also intensifying. The U.S. government introduced a new coordination framework called "Gold Eagle" to increase detection and prioritisation of critical infrastructure vulnerabilities and speed up patching.

· White House to launch "Gold Eagle" critical infrastructure vulnerability response framework

The security industry is also accelerating efforts to reshape product portfolios around AI.

Sophos launched its security platform "Sophos Fusion", which integrates security operations with endpoint, network, account, email and cloud protection. Sophos stressed that Fusion is an AI-native cyber defence system. Aton joined the initiative "Project Canopy" as a Defending Partner to expand AI-based vulnerability defence technology. Bespin Global, a company specialising in AI services and solutions, launched "HelpNow AI SOC", an AI-based security operations centre service. Offensive cybersecurity firm Theori launched "Xint Pulse", an on-demand diagnostic solution based on its AI-based vulnerability detection platform "Xint".

· Sophos launches AI-native defence system "Fusion" · Aton joins AI-based vulnerability defence initiative "Project Canopy" · Bespin Global launches AI security operations centre "HelpNow AI SOC" · Theori launches on-demand AI penetration testing solution "Xint Pulse"

Other moves and issues involving companies in and outside South Korea in the security sector were also compiled.

AWS is expanding its multicloud security business by extending the protection scope of Security Hub to Microsoft Azure. OpenAI unveiled an internal AI system called "GPT-Red" that attacks its own AI models to find prompt injection vulnerabilities. The system focuses on finding vulnerabilities before they reach users. Perplexity unveiled "Space", a sandbox platform to improve security and long-running task performance for its AI agent service "Computer". Global open-source solution companies Red Hat and IBM will formally launch "Lightwell", which provides automated vulnerability response capabilities.

· AWS Security Hub expands protection to Azure, adds new features · OpenAI unveils internal AI "GPT-Red" to catch prompt injection · Perplexity launches security sandbox "Space" for AI agents · Red Hat and IBM launch "Lightwell" automated vulnerability response service

Identity management startup Oak, which blocks unauthorised access to corporate applications, raised $60 million in seed funding. Oak provides a platform that automatically identifies access permissions within a company. It supports finding potential risks based on identifying each employee's application accounts and AI agents.

· Oak raises $60 million seed round, targets management of corporate app and AI agent permissions

Winstecnet is targeting the market for 대응 to the National Network Security Framework (N2SF) by promoting its recently launched generative AI data security solution, SNIPER AIVAX. AI security company Aim Intelligence and information security company Pado (PADO) signed an MOU for joint technology development to target the AI cybersecurity market. API governance company Softfreak supplied its automated vulnerability diagnosis solution "F-Avata" to the Korea Medical Device Safety Information Institute. AI digital forensics company Uraq will target the Indian market after securing a proof-of-concept project from major investigation and forensic science agencies in India. ITCEN PNS formed partnerships with U.S. post-quantum cryptography (PQC) specialist American Binary and digital twin solution provider RCK to expand PQC business in and outside South Korea.

· Winstecnet targets N2SF response market with AI data security solution · Aim Intelligence cooperates with Pado, says "building an AI security ecosystem centred on white-hat hackers" · Softfreak supplies automated vulnerability diagnosis solution to Korea Medical Device Safety Information Institute · Uraq targets Indian market with AI forensics, secures proof-of-concept project · ITCEN PNS cooperates with American Binary and RCK on PQC business

AI security company SentinelOne and MDR specialist Pagonetworks held a press briefing on the 16th and launched an AI-based joint security operations service called "Hybrid MDR". LG Uplus disclosed last year's information security activity results, including AI-based security monitoring and upgrades to its personal data protection framework. KT launched its "public AI email security" service after obtaining cloud service security certification (CSAP) from the Korea Internet & Security Agency (KISA).

· "AI attacks are blocked by AI" as SentinelOne and Pagonetworks launch "Hybrid MDR" · LG Uplus publishes "Information Security White Paper 2025", discloses AI security capabilities · KT launches "AI email security" for public sector after obtaining CSAP certification

KISA signed an MOU with Korea Water Resources Corp to develop and spread closed-circuit television (CCTV) technology specialised for construction sites.

· KISA and Korea Water Resources Corp to cooperate on CCTV security for construction sites

The fallout from a personal data leak incident at Tving is spreading to telecom carriers and platforms. After customers who received Tving subscriptions as compensation for telecom hacking 피해 again suffered a personal data leak, it was also confirmed that users who used Tving via platform simple logins were included among those affected. That has prompted calls for fundamental measures.

· Tving hacking fallout spreads, leaving telecom and platform users feeling wronged