AI Security

AI & ML 4 min A Chinese AI Hacking Tool and Claude Code Were Behind Korea's Bank Breaches, CrowdStrike Says Several South Korean lenders lost customer data in late September. The attacker's exposed servers revealed an... MA Maya Lin 12 Oct AI & ML 5 min Microsoft Says AI Is Finding Software Flaws Faster Than Anyone Can Fix Them Its annual threat report expects a record 72,000 tracked vulnerabilities this year, exploits within a day of d... MA Maya Lin 2 Oct AI & ML 5 min Google's Newest Model Went First to Vetted Defenders, With Its Cyber Guardrails Off Gemini 4 Argon is available only through the Fairwind programme for now. Most of its security claims come with... PR Priya Nair 2 Oct AI & ML 5 min Codex Read a Repository and Ran Its Owner's Code Two escapes from OpenAI's coding agent were reported in August and fixed within eight days. The more serious o... MA Maya Lin 22 Sep Cyber Threat Intel 6 min AI Agents Compromised 440 PaperCut Servers. Only Twelve Became Domain Admin. Eleven organisations fell in 26 seconds, and schools took 204 of the 395 victims. The gap between landing on a... PR Priya Nair 16 Sep Cybersecurity 6 min An Internal OpenAI Service Let One ChatGPT Account Read Another's Gmail Check Point found ChatGPT's isolated sandboxes could pass data through an internal package service — the same... KE Kenji Tanaka 14 Sep Cybersecurity 6 min Microsoft Has Patched 2,760 Flaws This Year. The Old Record Was 1,139. Microsoft shipped 974 fixes in September, a monthly record. The year is the number that matters: 2,760 so far... KE Kenji Tanaka 14 Sep Cybersecurity 6 min Cisco Found Seven Flaws in Every IOS XR Release, and Says AI Models Helped Two at CVSS 9.8, unauthenticated and network-reachable, none known to be exploited. The advisory names frontie... KE Kenji Tanaka 7 Sep Cybersecurity 7 min Three Labs Shipped Cyber Models in One Week, and All Three Disclosed Escapes Google, OpenAI and Anthropic each gated a cyber-capable model behind an application process. All three also di... KE Kenji Tanaka 4 Sep Cybersecurity 6 min CrowdStrike Built an Offensive AI Model and Named It Red Tempest CrowdStrike has shipped two models built with Nvidia: one that attacks a replica of your network, one that def... PR Priya Nair 3 Sep Cyber Threat Intel 6 min Google Fixed More Chrome Bugs in June Than in Two Years. Now Comes the Argument. Google patched 1,072 Chrome bugs in June against 1,036 in the prior two years. Whether that drains the exploit... PR Priya Nair 1 Sep Cybersecurity 8 min A Web Page Can Make Grok Decrypt Its Own Attack Instructions Encrypted commands sit on a page in plain sight. A filter cannot read them because reading them means running... KE Kenji Tanaka 26 Aug Cybersecurity 7 min Microsoft Took Eight Months to Patch a One-Click Copilot Data Theft Chain Varonis reported the chain in December 2025 and the fix shipped on 18 August 2026. In between, one link could... KE Kenji Tanaka 24 Aug AI & ML 11 min Microsoft Credits AI With Finding Bugs in May 2026 Patch Tuesday — and the Industry Should Pay Attention Microsoft's May 2026 Patch Tuesday addressed 118 to 138 CVEs across Windows, Office and Azure — and credited A... AI AI Tools Desk 23 May Cybersecurity 8 min Anthropic's Project Glasswing Lines Up AWS, Apple, Google, Microsoft, JPMorgan to Test Mythos for Cyber Defence Project Glasswing pairs Anthropic's restricted Mythos model with a roster of hyperscalers, banks and the Linux... CY Cyber Team 19 May Cybersecurity 7 min EchoLeak: How a Zero-Click Prompt Injection in Microsoft 365 Copilot Exfiltrated Enterprise Data EchoLeak (CVE-2025-32711), fixed by Microsoft in 2025, showed a malicious email could make Microsoft 365 Copil... CY Cyber Team 19 May Cybersecurity 5 min Anthropic's Project Glasswing Found a 27-Year-Old Bug in OpenBSD — Before Hackers Could Anthropic's Project Glasswing used an unreleased frontier model to find thousands of high-severity vulnerabili... CY Cyber Team 5 May Cybersecurity 5 min AI Is Making Phishing Emails Harder to Tell Apart From Legitimate Ones Large language models let attackers write fluent, personalised phishing at scale. Here is what employees and b... CY Cyber Team 5 May