Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

AI is making social engineering attacks significantly more effective, according to a new report from cyber insurance firm Resilience. These attacks were behind more than 85% of losses in the first half of 2026, compared to less than 20% during H1 2024. “Losses tied to phishing, social engineering, and transfer fraud have climbed from 17.7% of incurred losses in H1 2024 to 85.3% in H1 2026, the single largest increase in the report’s five half-year comparison,” Resilience says.

A study has found that AI chatbots can be more effective at social engineering than human scammers, WIRED reports. The researchers looked at a form of romance scam commonly known as “pig butchering,” in which scammers spend weeks or months building a relationship with the victim before tricking them into sending money for a phony investment scheme.

Enterprise security teams need to secure a risk surface that is constantly changing. However, the tools in their stack were built to check only a fraction of that risk. For confirmation, they rely on static snapshots and annual attestations. I now see this as the defining problem in GRC. When 451 Research (S&P Global) initiated coverage of TrustCloud in this space, they described a clear and growing divide.

Agent observability tools capture traces, metrics, logs, and evaluations across AI agent workflows. They help teams reconstruct execution paths, inspect tool calls and handoffs, diagnose failures, and monitor latency, cost, and output quality. This guide also covers complementary security platforms that discover agents, enforce runtime policies, or control the privileges agents receive. Agents don’t fail in straight lines.

A cyber loss model produces one distribution. How that distribution gets drawn changes what a reader can see in it, and the conventional projection hides the part most decisions depend on. ‍ The two views below contain identical data. One of them is close to unreadable for anything except the extreme tail, and the difference is worth understanding before the next time somebody asks what the number means. ‍

Human oversight is the only control in an AI program that can stop working while producing exactly the same evidence as when it worked. A failed encryption control throws errors. A monitoring pipeline that breaks stops delivering alerts. A review step that has become a formality still generates approvals, timestamps and sign-offs, and the compliance file looks identical. ‍ The asymmetry makes the design question secondary to the measurement one.

A SOC analyst starts a shift with a queue full of alerts. The first few investigations reveal repeated authentication failures from the same service account, blocked network scans from a known internal scanner, and endpoint events that three different sensors reported separately. Somewhere in that queue may be a real compromise, but the analyst has to work through the noise before finding it.

PoshC2 version 6.0, an open-source command and control framework, is notable for its robust capabilities in managing compromised hosts. Accompanying its release, a comprehensive list of Indicators of Compromise (IoCs) and a dedicated GitHub repository have been provided. These resources are designed to assist cybersecurity teams in detecting PoshC2, especially when deployed with its default settings, which less sophisticated attackers often utilize.