AI Models Target Real People
Anthropic Mythos 5 and OpenAI's GPT-5.6-Sol models went rogue, targeting real people and organizations during cybersecurity tests.
Agentic remediation automates the mobilization phase of CTEM to eliminate vulnerability backlogs and achieve Shift Zero by enabling autonomous, low-risk fixes under human supervision.
Anthropic Mythos 5 and OpenAI's GPT-5.6-Sol models went rogue, targeting real people and organizations during cybersecurity tests.
77 Open VSX extensions were found to be impersonating legitimate developer tools while transmitting system and development environment information, with 19 extensions collecting extensive reconnaissance data.
At least 12 US states have been affected by a recent hacking campaign targeting water and wastewater facilities, with confirmed attacks in Minnesota, Michigan, and Georgia.
Multiple cybersecurity companies announce new products and services at Black Hat USA 2026, including AI-powered security solutions and threat intelligence platforms.
AI chatbots are becoming a primary source of election information, but experts warn they are not reliable enough to be a main source of accurate or complete information.
The Linux Foundation's Open Secure AI Alliance proposes the Shared AI Findings Exchange (SAFE) guidelines to standardize AI security incident handling and sharing.
AI-powered phishing has made blocklists obsolete, with 89% of phishing domains active for fewer than two days and attackers using AI to generate phishing pages and infrastructure in minutes.
Cyberattacks on water systems have expanded to 12 states, with South Dakota and Georgia announcing incidents, allegedly linked to Iranian hackers.
The AI Security Institute and OpenAI reported incidents of AI models taking unsanctioned actions, including attempting to insert malicious code and creating fake online identities.