A clearer view of today.

A widespread credential-dumping incident, dubbed FortiBleed, reportedly exposed configuration artifacts for nearly 74,000 Fortinet devices, including usernames, email addresses, and plaintext admin passwords. NCSC UK also issued guidance after observing global targeting of Fortinet firewalls and VPN gateways, underscoring how quickly stolen credentials translate into real compromise attempts. The practical takeaway: treat exposed credentials as an active threat, rotate immediately, and verify remote administration paths are properly restricted and monitored.

Source: Help Net Security

Microsoft confirmed RoguePlanet, a Defender-related privilege escalation issue tracked as CVE-2026-50656, and said a fix is under development. The flaw is described as a race condition that can allow attackers to elevate privileges to the highest level through the Microsoft Malware Protection Engine. Organizations using Microsoft Defender should prioritize internal risk review, watch for mitigations/workarounds from Microsoft, and ensure detection controls are tuned for related suspicious behavior.

Source: Malwarebytes Blog

F5 released urgent patches for two critical NGINX flaws that could allow remote unauthenticated attackers to trigger memory corruption and potentially achieve denial of service or code execution. Because these are described as exploitable without authentication, defenders should assume scanning attempts may follow quickly after disclosure. Patch cadence matters: prioritize affected NGINX deployments, validate version numbers, and confirm compensating controls (WAF/allowlists/rate limits) are in place while updates roll out.

Source: SecurityWeek

International law enforcement action under Operation Endgame took down 106 SocGholish servers/domains and helped remediate nearly 15,000 compromised websites that were used to deliver fake software updates. The takedown targets an infection chain that historically provides initial access for downstream criminal activity. While the disruption is significant, defenders should treat it as a reminder to verify web and update mechanisms are hardened, integrity-checked, and that legacy persistence indicators are actively hunted.

Source: Help Net Security

ESET’s investigation describes “Gentlemen” as maintaining a framework of EDR-killing tools supplied directly to ransomware affiliates. Rather than relying on affiliate-side improvisation, the operators develop and distribute repeatable disablement tooling, increasing the reliability of hostile outcomes. For incident response and hardening, this elevates the importance of tamper protection, allowlisting, recovery-tested backups, and visibility into endpoint security process manipulation.

Source: ESET Blog

Research highlighted multiple enterprise AI breakages driven less by novel malware and more by broken trust boundaries—where URLs, agent tooling, and gateway components effectively become exfiltration or privilege-escalation paths. Examples include an issue in Microsoft 365 Copilot Enterprise Search enabling mailbox search and data exfiltration via indirect flows, and separate findings showing LiteLLM gateway authorization weaknesses that could expose provider credentials. The key operational lesson is governance-by-design: inventory AI tools and gateways, enforce least privilege, restrict network egress paths, and verify vendor fixes/configuration rather than assuming “approved” interfaces are safe.

Source: VentureBeat