27 articles that week
The Register, Saturday, October 24th, 2026
Symantec captured an attack agent built to hijack exposed AI endpoints and run up to 100,000 attack steps without human approval.
more →
Dark Reading, Friday, October 23rd, 2026
Attackers now impersonate customers and brands, calling help desks for resets instead of using malware; experts explain how to close those doors.
more →
Help Net Security, Friday, October 23rd, 2026
Huntress tracked an attacker replaying stolen Microsoft 365 session tokens from AiTM and device-code phishing to bypass MFA at 79 organizations.
more →
TechTarget, Friday, October 23rd, 2026
41% of organizations faced a deepfake audio social engineering attempt; help desks need verification that works however good the fake is.
more →
Dark Reading, Friday, October 23rd, 2026
Insurers are moving beyond questionnaires to continuous telemetry and in-house security services as AI and cloud risks outpace loss data.
more → TechTarget, Friday, October 23rd, 2026
LMG Security's Sherri Davidoff explains coverage types, hidden exclusions and contingent business interruption pitfalls in cyber policies.
more → Infosecurity Magazine, Friday, October 23rd, 2026
Kroll finds 69% of breach victims were hit through a third party, and 79% of those through a compromised technology company.
more →
Infosecurity Magazine, Friday, October 23rd, 2026
A Hogan Lovells partner argues AI-driven discovery reveals that severity scores were never the same thing as business risk.
more → Dark Reading, Friday, October 23rd, 2026
Microsoft's Guardian Agents treat AI agents as insiders with their own identities and risk scores; analysts doubt agents can police agents.
more → Computer Weekly, Friday, October 23rd, 2026
A Dragos analysis of the Monterrey water utility case shows AI shrinking the time from IT foothold to OT attempts, not inventing new OT attacks.
more →
CSO, Thursday, October 22nd, 2026
Rapid7 data shows attackers concentrate on a narrow set of critical flaws, so defenders should prioritize that profile over raw CVE counts.
more →
TechTarget, Thursday, October 22nd, 2026
How to use just-in-time access to cut standing privilege, where it fits with PAM and zero trust, and when not to deploy it.
more → The Register, Thursday, October 22nd, 2026
A pen tester found 'password' as an edge firewall's enable password; the hour-long fight that followed was really about IT bonuses.
more → Help Net Security, Thursday, October 22nd, 2026
Cisco Talos shows attackers planting callback numbers in real Dell and Staples order emails that pass DKIM and DMARC, plus other platform abuse.
more → Dark Reading, Thursday, October 22nd, 2026
Specialty carriers now sell AI coverage while major insurers add AI exclusions, so firms should review policies before year-end renewals.
more → Help Net Security, Thursday, October 22nd, 2026
Anthropic's red team found unnecessary exposure, weakening config combos and pre-auth code paths across 25 open-source projects.
more → Cybersecurity Dive, Thursday, October 22nd, 2026
CrowdStrike documents threat actors running agentic AI campaigns and finds token budgets, not engineering skill, limit AI-run intrusions.
more →
SC Media, Wednesday, October 21st, 2026
Explains how continuous threat exposure management differs from vulnerability management and five signs a program hasn't made the shift.
more →
Help Net Security, Tuesday, October 20th, 2026
95% of IT leaders see frontline workers bypassing slow logins by sharing credentials or staying signed in; 62% saw recovery abuse attempts.
more → SC Media, Tuesday, October 20th, 2026
Action1's field CTO says new CISA, NSA and Commerce guidance on AI developer security is a warning the industry should not ignore.
more → Help Net Security, Tuesday, October 20th, 2026
Malicious calendar invites carrying phishing links have surged, and invites can remain on calendars even after the email is quarantined.
more → Help Net Security, Monday, October 19th, 2026
A Microsoft and NVIDIA paper warns that firms running open-weight models must supply the guardrails and monitoring a provider would otherwise handle.
more →
Dark Reading, Monday, October 19th, 2026
An Omdia analyst argues the industry is conflating vulnerability volume with risk and should talk to boards about resilience and business impact.
more → The Register, Monday, October 19th, 2026
Terra Security finds GitHub security fixes later reported broken within 90 days jumped from 0.2% to 2.3% in 2026.
more → Help Net Security, Monday, October 19th, 2026
Semperis survey: attackers threatened staff families at 37% of ransomware victims, and swatting and doxxing are now part of the playbook.
more → Infosecurity Magazine, Monday, October 19th, 2026
Moody's says known exploited vulnerabilities take a median 87 days to patch, a gap that grows riskier as AI speeds exploitation.
more → TechTarget, Monday, October 19th, 2026
A practical look at where AI helps SOC teams, from alert triage to investigation summaries, with implementation caveats for each.
more →