Cybersecurity briefing, 2026-09-07

An attack technique built to fool AI is now smuggling phishing lures past email filters that read words.

• MikroTik SSH Flaws Draw Active Exploitation
• ScreenConnect Clients Are Spreading Scripts Across Hosts
• Google’s Sixth Chrome Zero-Day Hits V8 Again
• Sansec Says Magento Zero-Day Is Backdooring Stores
• TeamCity Breach Reached JetBrains Cadence Backup
• JSCeal Replays Google Sessions After Malvertising
• Microsoft Sees Unicode Smuggling in Phishing Campaign
• FalconFlank’s Public PoC Puts CrowdStrike at Risk
• Wordfence Sees Mass WordPress Shell Drops
• NetScaler Probes Rise After PoC Disclosure
• N-able Hotfix Churn Leaves N-central Exposed
• Rhysida Moves Berlin Theft Into Auction Mode
• Frontier Models Cross Into Autonomous Exploit Work
• CISA and G7 Push PQC Into Procurement
• VMware Patches Guest-to-Host Bugs as vCenter Fires Burn
• ShipMonk Breach Broke Trezor's Deletion Assumption
• Dropbox Warns of Lenovo Verification Abuse
• OpenAI Recasts Agent Wiki Abuse as Disclosure Gap