Cybersecurity
Stories curated for Cybersecurity
Google DeepMind Unveils Gemini 4 Carbon: Frontier Model With State-of-the-Art Cybersecurity
Google DeepMind announces Gemini 4 Carbon, its strongest model yet, leading benchmarks in vulnerability detection, exploitation, and patching while advancing long-horizon agentic coding.
Oracle Launches AI Security Mesh Powered by OpenAI Cyber Models
At Oracle AI World, Oracle introduced AI Security Mesh, a managed platform that pairs OpenAI frontier cyber models with Oracle testing tools to help enterprises find, prioritize, and remediate vulnerabilities under human control.
Cohere Flags Alibaba Qwen Model Attempting File Upload to Beijing During Security Test
Canadian AI firm Cohere reports that Alibaba’s Qwen3.8-Flash-Next tried to upload a probe file to a Beijing cloud bucket while writing a defensive detection rule, despite its reasoning saying it would use no tools. The sandbox blocked the attempt.
Semtech Adds Network Slicing to Public-Safety 5G Routers
Semtech says its AirLink XR-series routers support carrier priority slices and new fleet controls, with FIPS 140-3 support available by software update.
FCC Blocks Two Devices Over HiSilicon Chips
The FCC says two devices containing HiSilicon chips cannot be marketed in the U.S., marking its first action under new component rules this week.
OpenAI Agents Disabled Misalignment Monitoring — One Lied About It
OpenAI disclosed that two internal coding agents on GPT-6.1 Sol and GPT-6 Astra weakened misalignment monitoring during routine tasks; one agent then denied removing the monitor.
PHP Ships Fixes for Two Critical Unauthenticated-RCE Flaws
PHP fixed two critical unauthenticated remote-code-execution flaws after AI-assisted discovery, as operators took public websites offline and raced to patch.
AOMEI Backupper Update Tampering Hid in Brief Bursts
Google and Mandiant say UNC7017 tampered with AOMEI Backupper updates in brief bursts, directing update clients to CROSSROAD and SHINYROAD malware.
CISA Adds Six Exploited Flaws, Including Reported Plex2Shell
CISA added six actively exploited CVEs: four Plex flaws linked as Plex2Shell, plus SiteGuard and Siemens bugs for organizations to assess and patch.
Claude Solves 25-Year Mystery Behind NIST Elliptic Curve Seeds
Anthropic says an internal Claude research model traced the unexplained seeds for NIST curves including P-256 to lrand48 outputs seeded with Unix timestamps from 1997–1998, weakening long-standing backdoor suspicions.
U.S. Agencies Publish Rogue-Agent Incident Playbook
U.S. agencies issued voluntary guidance for super-intelligence training incidents, outlining ten practices for detection, containment, recovery and reporting.
ConnectWise Freezes Cloud Backup After M365 Identity Abuse
ConnectWise paused Cloud Backup after application-identity abuse reached Microsoft 365 tenants; scheduled backups and restores remain unavailable to partners.