The First Government Breach by an AI Agent Won't be the Last
Last week I wrote a piece talking about OpenAI’s misalignment blog and how it was a flashing red neon sign for security teams.
Security | Threat Detection | Cyberattacks | DevSecOps | Compliance
The Latest Cybersecurity News & Insights From Around The World
The latest News and Information on Security Incident and Event Management.
The New CISO Ep. 151 - Sean Murphy | Complacency Kills: Why More Discomfort Might Fix Your Burnout
Sean Murphy spent more than twenty years in the CISO chair and walked away from it while things were going well. In this episode of The New CISO, he returns to talk with Steve Moore about trading the operational seat for a field CISO role at F5 — and why getting too good at the job was the warning sign.
Can AI agents replace SOC analysts?
No two tech stacks are the same, so no agent can protect every organization the same way. This Short explains why human judgment needs to stay in the loop when it comes to agentic AI in security operations and what it means to build AI that handles the heavy lifting without sacrificing analyst control.
Extend Investigations with Falcon Next-Gen SIEM Federated Search and Snowflake
See how CrowdStrike Falcon Next-Gen SIEM Federated Search extends investigations to security data stored in Snowflake without requiring the entire dataset to be ingested into the SIEM. In this demo, an analyst pivots from suspicious endpoint activity in Falcon to Corelight network telemetry retained in Snowflake. Using remoteTable(), the analyst searches the remote data directly from Advanced Event Search, retrieving only the context needed to investigate the affected host and understand the broader scope of the incident.
Try Sumo Logic in minutes: see SIEM and Dojo AI agents in action
You already know the feeling. An alert fires, and you’re the one digging through logs to figure out if it matters. A query takes three tries to get right. A tool demo looked great, but you still cannot picture it running against your own environment. Before dedicating too much of your over-committed schedule to a proof of concept, you want to know one thing: does this actually work the way they say it does?
What OpenAI's Misalignment Reports Tell Security Teams About AI Agents
AI Agents are no longer the new insider threat. They’ve become firmly planted as the primary challenge for security leaders when it comes to insider threat activity. They have credentials, they act autonomously, and increasingly, they can take real actions on real systems. How would you know when one of your digital workers starts doing something it’s never done before?
Axoflow Launches AxoDetect, Bringing Detection Into the Pipeline and Making the SIEM Optional
Now in early access, AxoDetect runs Sigma rules in stream - alerts travel to the SIEM, and full-fidelity logs land in AxoLake, a low-cost security data lake.
CISA's Logging Reference Architecture for OMB M-26-14: What federal agencies should do next
On August 20, 2026, the Cybersecurity and Infrastructure Security Agency (CISA) published its Logging Reference Architecture (LRA), the implementation guidance federal civilian executive branch (FCEB) agencies have been waiting for since the Office of Management and Budget's Memorandum M-26-14 reset the requirements for enterprise logging.
Sumo Logic MCP server: bringing SIEM and log data into Claude and other AI clients
More security and operations work now happens inside an AI client instead of a dedicated console. That shift creates a gap for any platform that isn’t part of the conversation. Every time an analyst needs to triage an insight or check a log, they have to leave the AI client and go open a different tool. Sumo Logic closes that gap with a Model Context Protocol (MCP) server.
The only perfect Endpoint Prevention and Response (EPR) score in 2026 belongs to Elastic
Elastic sits at the very top of this year’s AV-Comparatives' CyberRisk Quadrant within the 2026 Endpoint Prevention and Response (EPR) test with the only protection scores at 100%, combined with both the lowest modelled operational footprint of any tested product and zero false alerts.