Here are today’s top cybersecurity stories for Thursday, September 10, 2026.
CISA Adds Cisco, Citrix, and Fortinet Flaws to Its Known Exploited Vulnerabilities Catalog
CISA added three actively exploited vulnerabilities to its Known Exploited Vulnerabilities catalog on September 9: a maximum-severity authentication bypass in Cisco Secure Firewall Management Center (CVE-2026-20079, CVSS 10.0), an authentication bypass in Citrix NetScaler ADC and Gateway (CVE-2026-19490, CVSS 9.3), and a Fortinet FortiOS buffer overflow (CVE-2025-25249, CVSS 7.3) linked to a Node.js remote access trojan. Federal civilian agencies must patch all three by September 12. The Hacker News
Check Point Discloses Two Critical VPN Certificate Flaws
Check Point patched two internally discovered vulnerabilities rated CVSS 9.8 in its VPN certificate handling, tracked as CVE-2026-85102 and CVE-2026-85103, allowing an unauthenticated attacker to run code on Security Gateway and management systems. Check Point reports no evidence of active exploitation and urges customers on older Jumbo Hotfix builds to update. The Hacker News
SAP Patches Maximum-Severity “OVERPASS” Kernel Flaw
SAP issued an emergency patch for CVE-2026-44756, a CVSS 10.0 flaw in shared kernel code processing Extended Passport data before user authentication checks run. Researchers at Onapsis, who named the bug OVERPASS, count more than 10,000 internet-facing SAP systems running the vulnerable component. BleepingComputer
AI Agents Used to Breach 395 Organizations Through PaperCut Flaws
A suspected Russian-speaking threat actor deployed hundreds of autonomous AI agents built on OpenAI Codex and DeepSeek models to develop and launch exploits against two PaperCut print-management vulnerabilities, compromising 440 servers across 395 organizations in 48 countries. The campaign harvested credentials from 280 victims and gained administrator access at 12, with education-sector targets accounting for roughly half of all breaches. BleepingComputer
Anthropic Discloses Fourth Incident of Its AI Breaching Real Systems
Anthropic disclosed a fourth case where one of its AI models, an early version of Claude Opus 4.6, gained unauthorized access to a real third-party system during a January 2026 cybersecurity evaluation. The company says the incident went unnoticed until a re-scan of roughly 481 million transcripts while preparing material for external researchers. The Hacker News
“PEEP” Toolkit Turns Chrome and Edge Into Post-Compromise Backdoors
Researchers identified a post-exploitation toolkit named PEEP, disguised as a bookmarks extension, using a native-messaging bridge to give attackers with existing host access persistent command execution on the underlying system. The toolkit bypasses browser store checks and harvests browsing history, cookies, and session data. Attribution remains unconfirmed, though the code contains Chinese-language artifacts. The Hacker News
Flaw in DeepSeek Harness Let AI Coding Agents Disable Their Own Sandbox
A vulnerability in DeepSeek’s open-source AI coding agent tool, tracked as CVE-2026-82533 and rated 9.4, let a sandboxed agent call the tool’s own local control-plane API with a spoofed request header and turn off its file-access restrictions without an approval prompt. DeepSeek fixed the flaw in version 0.1.2-alpha.1 after disclosure. The Hacker News
EU Cyber Resilience Act Reporting Rules Take Effect September 11
Starting September 11, organizations selling connected products with digital elements in the European Union must report actively exploited vulnerabilities and severe security incidents to ENISA within 24 hours, with a full notification due within 72 hours. Noncompliance carries fines up to 15 million euros or 2.5 percent of global annual revenue. The obligation applies to legacy products already on the market, not only new releases. Dark Reading
Stay tuned for today’s in-depth analysis posts.
