Morning Brief

In what they call the first-ever real-world agent-to-agent exploitation method, Pillar Security researchers say they discovered an exploit in the repository behind Google's Agent Development Kit for Python that could allow attackers to compromise supply chains. In other words, now we know that one AI agent can be used to control and compromise another one that has more privileges. The security snafu existed in google/adk-python, an open source Python toolkit with more than 90 million downloads used to build and deploy AI agents. Google has since fixed the underlying issue in the repository…

CISA has warned that attackers are actively exploiting a critical authentication bypass vulnerability in N-able N-central. Tracked as CVE-2026-18577 , the flaw affects N-central servers running versions earlier than 2026.3.1.7. N-central is a remote monitoring and management platform widely used by managed service providers to administer customer systems. Because the platform provides centralized access to many endpoint devices, a compromise could enable attackers to move across managed environments. CVE-2026-18577 is classified as an authentication bypass vulnerability via an alternate path…

A new proof-of-concept reveals how attackers can turn Microsoft Copilot, the AI assistant embedded in Microsoft 365, into an unwitting accomplice for business email compromise (BEC) and large-scale wire fraud. The demonstration shows that a single compromised employee account can escalate, with alarming speed, into full CEO account takeover and the theft of a quarter of a million dollars, with minimal technical effort from the attacker. The attack begins the moment threat actors gain access to a regular employee’s inbox. Rather than relying on traditional “living off the land” techniques like…

Las Vegas, United States, August 4th, 2026, CyberNewswire As AI-assisted attackers compress exploitation timelines to hours, Mallory turns live adversary intelligence into prioritized, policy-governed action across the tools security teams already run Mallory , the AI-native Threat and Exposure Management platform, today introduced a unified context and intelligence layer for security teams. The architecture has three parts: a context graph that correlates attack surface, threat, and vulnerability data; an intelligent reasoning layer that determines what matters and why; and a policy and…

Cybersecurity researchers have disclosed details of an active, multi-wave campaign that employs social engineering lures themed around Adobe and Zoom software updates, business document reviews, and system maintenance utilities to stealthily deploy Remote Monitoring and Management (RMM) programs like ConnectWise ScreenConnect. The campaign has been codenamed SMOKE#SCREEN by Securonix Threat

Google deleted three AI agent workflows from its Agent Development Kit (ADK) Python repository. Pillar Security showed that a public GitHub issue could manipulate a triage agent into triggering a privileged code-fixing agent. The researchers said the public agent could be prompt-injected into posting /adk-issue-fix as adk-bot. They identified the bot as a collaborator, so that comment satisfied

If you want to bypass AI guardrails designed to stop models from assisting with cyberattacks, you often just have to ask the right way, according to researchers from Cisco Talos. Simply claiming you own the servers you're targeting or that you're taking part in a capture-the-flag or bug bounty exercise was often enough to persuade models to cooperate. Talos researchers have been poring over prompt logs and artifacts recovered from threat-actor endpoints running tools such as Claude Code, Codex, Cursor, and Gemini to learn how suspected threat actors are abusing LLMs. The big takeaway from…

CISA has added three new vulnerabilities to its Known Exploited Vulnerabilities (KEV) Catalog , based on evidence of active exploitation. CVE-2026-9198 IBM Langflow Code Injection Vulnerability CVE-2026-18556 N-able N-central Authentication Bypass Using an Alternate Path or Channel Vulnerability CVE-2026-34486 Apache Tomcat Missing Encryption of Sensitive Data Vulnerability These types of vulnerabilities are a frequent attack vector for malicious cyber actors and pose significant risks to the federal enterprise. Binding Operational Directive (BOD) 26-04: Prioritizing Security Updates Based on…

A new Russian loader-as-a-service named DOUBLECUP uses ClickFix attacks to hide malicious code in PNG images cached by victims' browsers, ultimately delivering CountLoader to Windows and macOS devices and a new remote access trojan named DeviceManager to Windows systems.