Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

PwC recently did an AI agent survey where they found the following: This all sounds great, right? For many reasons it is, but agentic AI creates a challenge of visibility for organizations into how AI agents are communicating with each other and external third-party vendors. Imagine a multitude of AI agents autonomously exchanging data across a complex mesh of third-party vendors and applications.

Malware, as one of many cyber threats, is not some random annoyance. Yet, there is nothing polite about it. It bypasses your firewall and establishes itself in your system. Then, escalated privileges are granted, and processes are killed. If you are particularly unlucky, malware encrypts your core and sticks around like a parasite in the CI/CD. So, it’s not about chaos but orchestration. That means you’re forgetting about something.

A newly surfaced set of vulnerabilities in the SonicWall SMA100 series appliances has captured the attention of cybersecurity professionals. While SonicWall has released patches for CVE-2025-40596 through CVE-2025-40599, and media reports point to a surge in Akira ransomware attacks targeting SonicWall SSL VPN infrastructure, CISA has not formally confirmed exploitation of these specific vulnerabilities by Akira at this time.

For years, security frameworks have served as essential tools for aligning cybersecurity practices, but they’ve also come with limitations. Designed primarily for compliance, many frameworks are rigid by nature, sometimes to the extent of being a checklist, making them ill-suited for today’s dynamic risk environments. But the threat landscape has evolved, and so too must our approach.

Every demo promises that the AI is smarter, faster, and more reliable than humans.

You’re an IT professional whose teams manage network and security. They face millions of alerts each day, from threat detections to access anomalies and from connection failures to latency spikes. Your costs are high: it’s been hard to scale your team to cope with the growing volume of events. But the stakes are also high. Miss the wrong operational alert and your network team gets swamped with tickets. Miss the wrong security alert and your whole business is exposed.

AI agents are being rapidly embedded into the SaaS ecosystem to streamline operations, trigger complex workflows, and interact with sensitive data and systems. From automating calendar updates to executing code and accessing cloud data stores, they are becoming integral to business processes. But with this integration comes risk. AI agents are often quickly deployed across SaaS environments by employees, without centralized tools to govern them.

A new release of CrowdStrike Falcon Adversary Intelligence delivers automatically prioritized threat intelligence tailored to each unique customer environment. By surfacing the right intelligence at the right time, this update enhances SOC workflows by enabling faster triage, deeper investigations, and more confident response. Today’s organizations understand threat intelligence is critical to stay ahead of adversaries, but many struggle to determine which threats matter most and how to act on them.

CrowdStrike is announcing a native integration between CrowdStrike Falcon Shield SaaS security and the OpenAI ChatGPT Enterprise Compliance API, adding visibility and security posture capabilities for mutual customers’ ChatGPT Enterprise environments. This integration helps security teams inventory and monitor AI agents across their organization — including who created them, what they access, and how they’re shared — so teams can consistently apply existing security controls.

As the Chief Information Security Officer at Snyk, my primary role is to ensure the security and integrity of our products, our systems, and our customers' data. But my responsibility extends beyond our walls. It involves championing a vision for a more secure digital world—a vision I am proud to say we share with the U.S. Cybersecurity and Infrastructure Security Agency (CISA).