Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

Today, Mend.io is launching Contextual Project Classification, an AI-native feature that automatically analyzes your codebase to identify which applications handle sensitive data like payments, healthcare records, and PII, enabling true risk-based security prioritization.

Anyone who knows me knows I’m passionate about mindfulness. Because I genuinely believe it makes us better humans. But also, because I have one of those brains that desperately needs it. I’m easily distracted and I start new ideas before finishing old ones. My attention can scatter in a hundred directions. I wrote before how I clicked on a phishing test because I was multitasking and running on autopilot. And that moment really changed the direction of my career and my research.

As I was writing my latest book, How AI and Quantum Impact Cyber Threats and Defenses, I was hit by how many theoretical and real attacks there are involving AI. There are attacks committed by AI and attacks committed agsinst AI, and I’m not sure which category is bigger.

AI security tools are no longer just defensive layers. They are high value targets being studied, fingerprinted, and bypassed much like traditional endpoint detection and response (EDR) platforms and antivirus solutions were in their early days. The speed and scale at which these tools are being deployed makes reactive defense increasingly unsustainable.

Your Bedrock agent running on EKS receives a prompt through your RAG pipeline. CloudTrail logs it as a normal bedrock:InvokeModel event—status 200, authorized IAM role, expected endpoint. But inside the container, the agent’s response triggers a tool call that spawns curl to an external IP, exfiltrating the context window. GuardDuty doesn’t flag it because the connection routes through a permitted VPC endpoint. You open your AWS console and see a healthy API call.

You’ve sat through three vendor demos this week. Vendor A showed you an AI-SPM dashboard with a pie chart of misconfigurations. Vendor B showed you a nearly identical dashboard with different branding and a slightly wider set of compliance frameworks. Vendor C showed you posture findings with an “AI workload” tag that wasn’t in their product last quarter.

There’s a growing acceptance that AI is no longer optional in security. That battle is largely won. The more interesting question — and the one I keep getting asked — is what we actually believe AI should be responsible for, and where human authority must ultimately sit.

Agentic AI promises to improve work processes in all domains and industries. R&D is no different. Recently, Cato R&D built an internal self-evolving pull request (PR) review agent that keeps reviewers in flow by commenting only on high-impact, high-confidence issues, validating every change against its spec from the PR and Jira, and learning continuously from developer feedback through long-term, episodic memory. What were the results?

On April 2, 1796, a full house packed the Drury Lane Theatre in London, eager to witness the first showing of a newly discovered Shakespeare play. The problem was that William Henry Ireland wrote the play, Vortigern, and the entire production was a hoax. Although there was some controversy before opening day, several experts reviewed the manuscript and supporting documents and confirmed that the play was a long-lost Shakespeare original.