Cybersecurity on Seven Lights https://sevenlights.org/tags/cybersecurity/ Recent content in Cybersecurity on Seven Lights Hugo en-us Sat, 07 Nov 2026 03:30:00 -0500 OpenAI rates new GPT-6 Cyber model 'critical' after finding hundreds of software flaws https://sevenlights.org/2026/11/07/openai-gpt6-cyber-model/ Sat, 07 Nov 2026 03:30:00 -0500 https://sevenlights.org/2026/11/07/openai-gpt6-cyber-model/ OpenAI said Thursday its security-specialized GPT-6 Cyber model found 81 Firefox flaws and built 948 working Linux kernel exploits in testing, leading it to rate the system ‘critical’ and restrict access. Password manager mSecure orders resets without explaining the breach https://sevenlights.org/2026/11/05/msecure-password-manager-breach/ Thu, 05 Nov 2026 03:50:00 -0500 https://sevenlights.org/2026/11/05/msecure-password-manager-breach/ mSeven Software told current and former mSecure customers to change passwords, reset two-factor authentication and remove passkeys after ‘suspicious activity’, reportedly an encrypted customer database theft. The company has not said what was taken. Binance postmortem details months-long North Korean operation behind record heist https://sevenlights.org/2026/11/04/binance-heist-postmortem-north-korea/ Wed, 04 Nov 2026 05:45:00 -0500 https://sevenlights.org/2026/11/04/binance-heist-postmortem-north-korea/ A postmortem published by Binance describes how a North Korean hacking unit worked for months, through a fake job offer, poisoned software and the exchange’s own internal AI tool, before stealing $6.34 billion on October 29. Leak Exposes 41,000 Homeland Security Workers, Including Undercover Agents https://sevenlights.org/2026/10/30/dhs-leak-41000-workers-ice-agents/ Fri, 30 Oct 2026 03:50:00 -0400 https://sevenlights.org/2026/10/30/dhs-leak-41000-workers-ice-agents/ Records identifying more than 41,000 Homeland Security employees and contractors, among them undercover ICE agents, were published online in a dump dubbed IcePick, according to WIRED, Gizmodo, Ars Technica and TechRadar. Google Unveils Gemini 4 Carbon, a Frontier Model Aimed at Cyber Defence https://sevenlights.org/2026/10/30/google-gemini-4-carbon-cybersecurity/ Fri, 30 Oct 2026 03:49:00 -0400 https://sevenlights.org/2026/10/30/google-gemini-4-carbon-cybersecurity/ Google announced Gemini 4 Carbon, a frontier AI model it says leads rivals at finding, exploiting and patching software vulnerabilities. Access is restricted to approved defenders through the Fairwind Program.