Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

The EU Cyber Resilience Act (CRA) introduces a 24-hour reporting requirement for actively exploited vulnerabilities from September 11, 2026. For companies selling products with digital elements into the EU, meeting that deadline will require more than compliance documentation — it demands fast vulnerability identification, clear ownership, reliable SBOM visibility, and a remediation process that can move quickly.

Periodic vulnerability scans are no longer a reliable security program for hybrid environments. NIST's guidance treats scanning as a recurring, risk-based control that should account for changing vulnerabilities, historical results, authenticated coverage, and correlations between findings over time (NIST vulnerability-scanning guidance). Industry practice is moving in the same direction.

Network edge vulnerabilities are accelerating. Cato’s cloud-native architecture reduces customers’ attack surface and shortens exposure to emerging vulnerabilities. The Cato CVE Exposure Calculator shows what that difference could mean for your organization. Another Known Exploited Vulnerability (KEV). Another emergency CAB. Another weekend spent upgrading firmware. You have to respond. The question is how much each response costs the business.

AI coding assistants have gotten remarkably good at writing functional code. Syntax correctness rates are approaching 100%. Developers are more productive than ever. And yet the security picture tells a very different story.

Let me be direct about something the industry keeps dancing around: the vulnerability problem isn’t getting better. It’s getting structurally worse. The wave of AI-generated code and compression of time to exploit thanks to frontier AI models like Mythos is about to make “worse” look quaint. If your vulnerability and exposure management program is still built around scanner cycles, ticket queues, and CVSS scores, you’re not running a security program.

Imagine getting three separate security reports back for your new enterprise AI assistant: On paper, the application looks ready for production, but in reality, a threat actor bypasses your guardrails in minutes. How? By using the AI model as an intermediary. The attacker steers the LLM to invoke the internal utility tool, thereby bridging an untrusted prompt directly to the backend execution sink.

OWASP published version 1.0 of the Agentic Skills Top 10 on August 17, defining ten risk classes for the layer where agents find, load, and run reusable instructions and code. This standard arrived while the problem was still forming, mapping directly to AISVS, the Agentic Security Initiative Top 10, the MCP Top 10, ISO/IEC 42001, and the NIST AI RMF. It ships with working code for signing, sandboxing, and pinning.